C—H Bond Activation in Transition Metal Species from a

Jan 12, 2010 - The large body of experimental results allowed us to gain a better understanding of the C—H activation process. Even though ...... (3...
8 downloads 13 Views 2MB Size
Chem. Rev. 2010, 110, 749–823

749

CsH Bond Activation in Transition Metal Species from a Computational Perspective David Balcells, Eric Clot, and Odile Eisenstein* Institut Charles Gerhardt, Universite´ Montpellier 2, CNRS 5253, cc 1501, Place Euge`ne Bataillon, 34000 Montpellier, France Received September 16, 2009

Contents 1. Introduction 2. Mechanisms of CsH Bond Activation 2.1. Oxidative Addition 2.1.1. d 8 Configuration 2.1.2. d 6 Configuration 2.1.3. d 4 Configuration 2.2. σ-Bond Metathesis 2.3. 1,2 Addition 2.3.1. Early Transition Metals 2.3.2. Late Transition Metals 2.4. Beyond Classical Descriptions 2.5. Rebound Mechanism 3. Chemically Induced Selectivity 3.1. Thermodynamic Properties: Bond Energy Enthalpies Determination 3.2. Chelation-Assisted CsH Activation 3.2.1. ortho-CsH Activation 3.2.2. CsH Activation in N-Heterocycles 3.2.3. CsH Activation in Ancillary Ligands 3.3. CsH Activation under Molecular Recognition 3.4. Base-Assisted CsH Activation 4. Catalytic Alkane and Arene Functionalization 4.1. Alkane Dehydrogenation 4.2. Hydroarylation of Olefins 4.3. Borylation of Alkanes 5. CsH Oxidation 5.1. Shilov Chemistry: Methane Oxidation 5.2. Metal-Oxo Species 5.2.1. Inorganic Reagents 5.2.2. Non-heme Iron Biomimetics 5.2.3. Non-heme Iron Enzymes 5.2.4. Manganese Porphyrin Biomimetics 5.2.5. Copper Biomimetics 5.2.6. Copper Monooxygenases 5.2.7. Dehydrogenases 5.3. Metal-Hydroxo Species 5.3.1. Lipoxygenases 5.3.2. Manganese Porphyrin Biomimetics 5.4. Metal-Peroxo Species 5.4.1. Methyltrioxorhenium 5.5. Metal-Hydroperoxo Species 5.5.1. Activated Bleomycin

749 752 752 752 757 758 759 763 763 765 766 769 770 770 772 772 776 779 782 783 788 788 790 793 795 795 800 800 801 805 807 808 808 809 809 809 810 811 811 811 811

* To whom correspondence should be addressed. E-mail: odile.eisenstein@ univ-montp2.fr.

5.6. Bimetallic µ-Oxo Species 5.6.1. Copper Biomimetics 5.6.2. Methane Monooxygenase 5.7. Metal-Alkoxo Species 5.7.1. Galactose Oxidase 6. Conclusions 7. Acknowledgments 8. References

811 811 812 816 816 817 817 818

1. Introduction Alkanes constitute the cheapest and most abundant feedstock for organic chemicals. However, the redistribution of CsC and CsH bonds in hydrocarbons into more sophisticated value-added architectures is a highly difficult task. The bonds to be broken are thermodynamically strong and kinetically inert. Therefore, hydrocarbons are mainly used as fuels, and the energy content of the bonds is dispensed as heat. Consequently, the efficient and selective activation of CsH bonds, the elementary building block of hydrocarbons, constitutes an interesting alternative with potentially a huge economic impact. Homogeneous catalysis mediated by transition metal complexes is one of the most efficient ways to achieve both high activity and control of the selectivity in CsH activation. The Shilov system1 is an early example of an alkane oxidation catalyst that long remained an isolated case of metal-catalyzed CsH activation. It was generally thought that the hydrido-alkyl, product of the CsH activation, was too unstable to be observed. With the work of Bergman2 and Jones3 on the intermolecular CsH activation of aromatic and aliphatic CsH bonds by Cp*ML (Cp* ) η5-C5Me5; M ) Rh, L ) PMe3; M ) Ir, L ) CO, PMe3), it proved possible to isolate the CsH oxidative addition products. This paved the way to intense activity and resulted in the publication of many reviews on stoichiometric4–11 and catalytic12–21 CsH activation. The large body of experimental results allowed us to gain a better understanding of the CsH activation process. Even though CsH activation is to be contrasted with alkane functionalization involving replacement of CsH hydrogen by a functional group X, the latter often proceeds through CsH bond activation when catalyzed by transition metal species. Depending on the nature of the metal M and the ligand set Ln in the active LnM species, different situations are encountered, and the CsH activation processes are classified according to four main different mechanisms (Figure 1). Oxidative addition is the most common mechanism by which a CsH bond cleaves and a MsC bond and a MsH bond are formed. It is well accepted that this mechanism

10.1021/cr900315k  2010 American Chemical Society Published on Web 01/12/2010

750 Chemical Reviews, 2010, Vol. 110, No. 2

David Balcells was born in Sant Martı´ de Tous (Catalonia, Spain) in 1978. In 2001, he received the Bachelor of Science degree in Chemistry at the Universitat Auto`noma de Barcelona. From 2001 to 2006, he worked on a Ph.D. project under the guidance of Prof. Feliu Maseras, firstly at the Universitat Auto`noma de Barcelona (2001-2003) and afterwards at the Institute of Chemical Research of Catalonia (2004-2006). His Ph.D. focused on the application of computational chemistry to the study of the asymmetric synthesis of chiral sulfoxides. In 2006, he obtained the Ph.D. degree in chemistry. In 2007, he obtained a Sanofi-Aventis postdoctoral fellowship for 3 years in the group of Prof. Odile Eisenstein, where he worked in the study of green synthetic methods and CsH activation processes from a computational perspective. He has earned a Juan de la Cierva position for 3 years in the group of Prof. Agustı´ Lledo´s, where he will work on the theoretical study of metal-catalyzed water oxidation.

Eric Clot was born in 1967 in Marseilles, south of France. After graduating from the Ecole Normale Supérieure de Lyon, he obtained the Agrégation de Chimie. After a Ph.D. with Odile Eisenstein and Claude Leforestier in 1995 on the dynamics of metal polyhydrides, he entered the CNRS in 1996. He has been promoted to Research Director in 2007. He conducts his research in the group headed by Odile Eisenstein in Montpellier. He also lectures at Ecole Polytechnique, Palaiseau. His research interest focuses on the computational study of the reactivity of transition metal complexes, with specific interest in the activation of inert bonds and applications to organic reactions. His research is conducted in close collaboration with many experimentalists throughout the world. He is especially happy when they share his keen interest in rugby, a sport for which he is a coach in Ecole de Rugby du Pic Saint Loup.

starts by coordination of the CsH bond to the metal vacant site (Figure 1). This mechanism is typical for electron-rich, low-valent complexes of the late transition metals (Re, Fe, Ru, Os, Ir, Pt) for which the higher oxidation state of the metal in the product and the necessary change in geometry upon formation of the two new bonds are not energetically penalizing. For early transition metals (groups 3 and 4, lanthanides and actinides) with d0 electronic configuration, an oxidative process is not possible, and the preferred mechanism is σ-bond metathesis involving usually an alkyl

Balcells et al.

Odile Eisenstein was born in 1949 in Boulogne-Billancourt, near Paris. After a Ph.D. in 1977 with Nguyen Trong Anh and Lionel Salem at University of Paris-Sud, Orsay, she joined the CNRS. She was research associate with Jack Dunitz at the ETH, Zurich, and with Roald Hoffmann at Cornell. She was assistant professor at University of Michigan at Ann Arbor. She became Research Director and head of the group of Theoretical Chemistry in Orsay in 1986. She moved to Universite´ Montpellier 2 in 1996 where she is still. She has been invited professor in several universities such as University of York (U.K.), Indiana University, which gave her an honorary degree, and UC Berkeley, where she was Miller Professor. She has received several awards including the Lebel award from the SFC, the Langevin award from the French Academy of Science, the Silver Medal of the CNRS, the Frankland RSC award, and the ACS Organometallics award. She has a keen interest in scientific journals and has been on the board of NJC for many years. She is Chevalier de la Le´gion d’Honneur. She is a member of the International Academy of Quantum Molecular Science (IAQMS).

or a hydride complex (Figure 1). The essential aspect of the σ-bond metathesis mechanism is the concerted formation (MsC and C′sH) and breaking (MsC′ and CsH) of bonds at the transition state. A reactivity formally equivalent to σ-bond metathesis is observed for late- or post-transition metals (Pd2+, Pt2+ or Pt4+, Hg2+), usually in a strongly polar medium. The putative intermediate is formed by electrophilic attack of the metal. In the product, the hydrogen atom of the substrate has been formally substituted by the metal center acting only as a Lewis acid, hence the classification as electrophilic substitution. Finally, the CsH bond can add across an unsaturated MsX bond in a 1,2 addition (Figure 1). This mechanism resembles σ-bond metathesis except that the newly formed XsH bond does not lead to release of XH because the MsX σ-bond is still present in the product. The functional groups X leading to such reactivity are generally amido and alkylidene complexes of early to middle transition metals, but alkoxy and alkylidyne complexes have also been shown to activate CsH bonds.

Figure 1. Schematic representation of the different mechanisms for CsH activation.

CsH Bond Activation in Transition Metal Species

From a conceptual point of view, despite seemingly large differences among the various mechanisms, there are three different fundamental aspects to consider in CsH activation: • How is the CsH bond brought into the proximity of the metal? • How is the CsH bond cleaved in the metal coordination sphere? • What is the thermodynamic stability of the organometallic product featuring the new MsC bond? A better understanding of the factors affecting the three fundamental aspects mentioned above is useful to design more efficient systems to promote selectiVe CsH activation. Rationalization of the reactivity observed is usually based on qualitative concepts such as steric bulk, electronic influence of the ligands, acidity of the CsH bond to be broken, etc. These are valuable tools to build a qualitative picture, but they lack the required quantitative aspects to effectively delineate the subtle influences of the ligand set, the metal, and the substituents on the reactants. Computational studies of the structure and reactivity of organometallic complexes have now proven to be an essential tool in homogeneous catalysis.22–26 A wealth of comparative studies has now established density functional theory (DFT) as a reliable method to study organometallic reactivity.27,28 Since the seminal review by Niu and Hall,23 the majority of the computational studies published in the field have used DFT methods and in particular the hybrid B3LYP approach. Together with the increase in computing power, this has allowed us to model systems closer to the actual experimental situation. One essential asset of computational studies is that they give access to information difficult, if not impossible, to obtain directly from experiment. One key piece of information for the studies of reaction mechanisms, which is not available by any other method, is the characterization of the transition state (TS) on the potential energy surface (PES). The energy of the transition state allows the evaluation of an energy barrier (∆Eq or ∆Hq) or an activation barrier (∆Gq), which can be compared with experimental values from kinetic studies, when available.29 The electronic structure of the TS allows characterization in more detail of the mechanism of the reaction in terms of bonds that are broken and made. The reactants and products connected by a given TS are also easy to obtain from calculations giving access to the thermodynamics of the reaction. Calculations are also useful to locate intermediates, which may be too unstable to be observed, and to suggest modifications to the experimental systems to favor or disfavor the intermediate. Calculations have thus contributed to the discovery of new pathways for CsH activation. Emphasis is often put on the CsH bond-breaking step in a multistep process because it is assumed to be the most difficult transformation due to the inertness of the CsH bond. However, studying the CsH activation step thoroughly is not enough to get insight into catalytic selectiVe CsH functionalization. Computing the energy profile for the entire catalytic cycle is a daunting task but it leads to a better understanding of the relative activation barriers for the individual steps. It also allows us to tackle the essential aspect of selectivity by comparing catalytic cycles leading to different isomers. Computation of the full energy profile also gives access to the rate-determining step for the transformation of interest and enables us, in principle, to propose

Chemical Reviews, 2010, Vol. 110, No. 2 751

alterations of the catalyst and experimental conditions to improve activity and selectivity. The purpose of this review is to describe the computational work that has addressed CsH activation by transition metal complexes since the reviews by Hall, Dedieu, Frenking, and their respective co-workers published in 2000 in the Chemical ReViews special issue “Computational Transition Metal Chemistry” edited by Davidson.23–25 The focus of the present review is on homogeneous CsH activation of alkanes and arenes with particular emphasis on selectivity issues in activation and functionalization. More specifically, calculations of CsH activation on bare metals are not considered. These calculations have been essential in order to unravel the basic electronic features of the CsH cleavage process and have allowed extensive benchmarking of the computational methods.30,31 However, most of the complexes active in CsH activation and functionalization are surrounded by ligands, and the latter have a critical influence on the outcome of the reaction. Moreover, given the size of the systems considered in the calculations, the majority of the studies have been carried out within the framework of DFT. Therefore, we will not describe any methodological issues because our main interest is on the chemical concepts obtained from the calculations. When necessary, methodology issues will be mentioned for specific cases. A review including description of computational works on CsH activation by surface organometallic complexes has just appeared.32 The first part of the review (section 2) addresses the computational characterization of the mechanisms for CsH activation shown in Figure 1. The description follows the usual partitioning between the various mechanisms,33 oxidative addition, σ-bond metathesis, 1,2 addition, but the calculations show that such a distinction is often difficult to make and that many intermediate situations exist (section 2.4). In the context of selective CsH oxidation, it is also interesting to mention radical processes, particularly important in biological and biomimetic reactions (section 2.5).34 This latter transformation has been well studied computationally because it offers challenges both in the definition of the model system and in the choice of the computational methodology. In the second part (section 3), the emphasis is put on selectivity issues. Because the CsH bond is a poor Lewis base, different strategies have been designed to achieve selective activation. Substituents on the reactant may influence the thermodynamics of the CsH activation, thus affording a means to control the site of reaction. A substituent can also be used as a directing group to trigger the metal toward a specific position as in the Murai reaction.35 More generally, the metal fragment can be designed such as to bring the target CsH bond close to the metal by molecular recognition.36 Finally, selectivity can also be induced by an external partner such as an added base as in the concerted metalation-deprotonation (CMD) mechanism proposed by Fagnou,37 or the ambiphilic metal-ligand activation (AMLA) proposed by Davies and Macgregor.38 The third part (section 4) describes the computational studies of catalytic cycles for CsH functionalization processes. Despite intensive efforts, only a few examples of transformations based on CsH activation have been shown to work under catalytic conditions. We focus here on the three different classes, which have led to extensive computational works. At the beginning of the 1980s, Crabtree

752 Chemical Reviews, 2010, Vol. 110, No. 2

developed iridium-based catalysts that showed activity in alkane dehydrogenation.39 Introduction of PCP pincer type ligands improved the catalytic performance as illustrated by Goldman, Jensen, and Kaska19,40,41 and led to alkane metathesis by tandem alkane dehydrogenation and olefin metathesis.42 In terms of CsC bond formation based on CsH activation, a significant breakthrough by Murai and Kakiuchi occurred in 1993 with the Ru-catalyzed selective hydroarylation of olefins ortho to the carbonyl group on aromatic ketones.35 The transformation was later shown by Gunnoe and Periana to work without using an orienting group.43,44 Also important to CsC bond formation processes is the catalytic borylation of alkanes developed by Hartwig because the alkyl-boranes are important reagents in various organic reactions.45,46 The cases where only the CsH activation step of a catalytic cycle has been studied computationally are described in different parts of the review. In the fourth part (section 5), the computational studies on alkane oxidation by transition metal complexes are considered. Historically, the Pt-catalyzed formation of methanol from methane described by Shilov long remained as the sole example of catalytic homogeneous CsH alkane oxidation.47 Nevertheless, the use of a stoichiometric amount of oxidant (usually Pt(IV)) limits the practical use of this catalytic system. The situation was dramatically improved in the 1990s by Periana with the Hg- or Pt-catalyzed formation of methyl bisulfate from methanol in hot sulfuric acid.48,49 In the field of alkane oxidation, MO and MDX (X)H, O, OH, M, R) species have been shown to promote CsH activation, mostly through a radical pathway based on the rebound mechanism. These processes offer selectivity different from the one observed in Shilov-type chemistry. The focus here is limited to non-heme systems because other contributions in this issue address the computational studies on cytochrome P450.50,51

2. Mechanisms of CsH Bond Activation The breaking of a σ(CsH) bond in the coordination sphere of a transition metal complex results generally from two synergic transfers of electron density: σ-donation from the bonding σ(CsH) molecular orbital (MO) into a symmetryadapted vacant orbital on LnM and π-backdonation from an occupied dπ MO on LnM into the antibonding σ*(CsH) MO. Both factors concur to weaken the σ(CsH) bond leading eventually to bond cleavage. However, a vacant site needs to be created on the metal to allow coordination of the σ bond. The experimental observation of a σ-complex is challenging because the interaction between the metal and the CsH bond is usually weak.52–56 Computationally, such intermediates are often encountered prior to the CsH cleavage step. The backdonation from the metal is usually effective from a nonbonding dπ MO on LnM, but in principle, any high-lying occupied MO could be involved in electron transfer into σ*(CsH) if symmetry allows. The various mechanisms for CsH activation thus depend on the actual nature and magnitude of the two contributions to the transfer of electron density: σ-donation from σ(C-H) and π-backdonation into σ*(C-H).

2.1. Oxidative Addition The main characteristics of the oxidative addition mechanism are the concerted breaking of CsH and formation of MsC and MsH in the TS, the formal increase by 2 units

Balcells et al.

Figure 2. Schematic representation of the σCH2 and πCH2 MOs for alkanes.

of the oxidation state of the metal, and the change of geometry of the complex to accommodate the two new σ bonds. Efficient CsH activation through the oxidative addition mechanism results from the interplay between the various requirements and depends naturally on the electronic configuration of the metal in the fragment LnM. The computational studies on CsH activation by d10 Pd and Pt complexes, where oxidative addition is prevalent, have been reviewed by Dedieu up to 2000.25 Sakaki and Tilset have updated the literature to 2005 with no significant changes in the fundamental aspects of CsH activation by d10 transition metal fragments.26,57 The CsH activation by d10 metal complexes will not be reviewed further in the present contribution.

2.1.1. d 8 Configuration The first systems in which the hydrido-alkyl complex (product of intermolecular CsH activation) has been directly observed have the general formula Cp*ML (M ) Rh, Ir; L ) PMe3, CO).3,6 These 16-electron d8 unsaturated fragments are generated in situ thermally or photochemically from coordinatively saturated 18-electron precursors through reductive elimination of H2 (or RH) or ligand dissociation. In alkane or benzene solution, they immediately react with the solvent to yield the oxidative addition product Cp*M(L)(R)(H). Following the publication of the work of Bergman and Jones, several computational studies have examined CsH activation by CpML,58–65 and they are described in detail in the review by Niu and Hall.23 Recently, Xavier et al. have studied at the MP4(SDQ)//MP2 level the CsH activation of methane promoted by (η5-phospholyl)Rh(CO)2 and compared it with CpRh(CO)2.66 The work by Su and Chu on the reaction between CpM(PH3) (M ) Rh, Ir) and propane is the first attempt to address computationally the important aspect of selectivity in CsH activation.67 Alkanes are very poor Lewis bases, and the only MOs available for σ-donation are σCH2 and πCH2 (Figure 2). Six different structures (1-6) describe the various ways that the metal fragment CpM(PH3) interacts with the primary and secondary CsH bonds (Figure 3). The six different situations yielded local extrema structures on the PES, although only three are regular transition states with only one negative eigenvalue of the Hessian matrix. For the other cases, two or three negative eigenvalues are obtained, the largest one (in absolute value) being associated with CsH cleavage and the others corresponding to rotational motion of the groups attached to the carbon atom. Table 1 collects the energies of the various extrema relative to separated CpM(PH3) and propane. Note that for Rh, no first-order TS could be located at the secondary position; 1 corresponds to a second-order TS with ν ) 906i cm-1 for CsH cleavage, and ν′ ) 78.7i cm-1 for rotation of the sec-propyl fragment. For both Rh and Ir, CsH activation is kinetically preferred at the primary position by 2.1 kcal mol-1 (Rh) and 1.65 kcal mol-1 (Ir). Selectivity is larger for Rh in agreement with experimental observations. The structures 1-3 correspond

CsH Bond Activation in Transition Metal Species

Chemical Reviews, 2010, Vol. 110, No. 2 753

Figure 3. Geometries of interaction between CpM(PH3) and primary (top) or secondary (bottom) carbon atom of propane. Table 1. Energy (kcal mol-1) Relative to Separated Reactants of the Extrema Structures Corresponding to the Various Geometries of Attacks (Number of Negative Eigenvalues in the Hessian in Parentheses)a Rh 1° Rh 2° Ir 1° Ir 2°

1

2

3

4

5

6

M[σ(CH)]

M(H)(C)

-0.92 (2) 0.13 (2) -4.47 (2) -4.13 (1)

-1.97 (1) 2.40 (3) -5.78 (1) -0.69 (2)

1.21 (2) 2.61 (2) -1.04 (3) -0.38 (2)

0.68 (3) 4.40 (2) -2.39 (2) 0.59 (3)

0.85 (3) 2.85 (2) -2.88 (3) -2.01 (3)

-0.05 (3) 2.64 (3) -4.31 (3) -1.32 (3)

-10.9

-15.6

-6.56 -6.07

-34.6 -31.7

a The σ-complexes and hydrido-alkyl products are connected through the first-order TS. M 1° and M 2° correspond to a reaction occuring at the primary and secondary carbon, respectively.

to situations where a σCH2-type MO is used for σ-donation, while the structures 4-6 use a πCH2-type MO. Table 1 shows that sterics at the CsH bond to be activated play a crucial role. Structure 3, featuring Et or Me close to the metal, is significantly less stable than 1 and 2. Structures 4-6 all lie at higher energy than 1 and 2, but the difference between these two sets of structures is larger for Rh than for Ir. There is also an influence of the nature of the metal. Metals with less diffuse orbitals (Rh) increase the preference for a σCH2type mode of interaction with the least sterically encumbered position of the alkane. The metal with more diffuse orbitals (Ir) is less selective both for the geometry of interaction (σCH2 or πCH2) and for the position of activation (primary or secondary). The negative values of the energy of the first-order TS indicate that a molecular precursor was located on the PES. Three σ-complexes (σ-Rh1, σ-Ir1, σ-Ir2) were optimized, and their energies are given in Table 1 relative to separated reactants. The energies of the products of the reaction, CpM(PH3)(H)(C3H7), are also given in Table 1. The stability of the σ-complexes is not strong enough to compensate for the loss of entropy (mostly translational entropy) associated with a change of molecularity upon formation of the complexes. For rhodium, the absence of a σ-complex at the secondary position is in agreement with greater preference for activation at the primary position. For iridium, the energy difference between the two σ-complexes is only 0.5 kcal mol-1, in favor of σ-Ir1, a value lower than the energy difference between the two corresponding TSs. The selectivity for activation at the primary position is thus only partially present in the precursor complex. The main difference between Rh and Ir lies in the thermodynamics of the reaction. In both cases, the oxidative addition is exothermic with respect to the σ-complexes but only by 4.70 kcal mol-1 for Rh and by 28.04 and 25.63 kcal mol-1 for Ir, because bonds are stronger to third than to second row metals. Consequently, oxidative addition at Ir is not reversible, and there is no possibility to equilibrate the various products through formation of the σ-complex by CsH reductive elimination. The energy difference between the two TSs thus determines the product ratio, and the calculated value of 1.65 kcal mol-1 corresponds to a 1.9:1 ratio at room temperature between activation at the primary and secondary positions. In contrast, for Rh the oxidative addition is not exothermic enough to prevent equilibration

through the σ-complexes. Consequently, the product resulting from reaction at the primary carbon can be obtained, in agreement with the experimental observations. Another important aspect of selectivity is the competition between CsH bonds in different substrates. Thus, although the CsH bond is stronger in benzene than in methane (and in other alkanes), aromatic CsH activation is usually preferred over aliphatic CsH activation both kinetically and thermodynamically (see section 3.1). This competition was addressed computationally by Bi and co-workers with B3LYP calculations of the CsH activation of benzene and methane by CpRh(PMe3)(H)2.68 The reductive elimination of H2 from the dihydride is computed to be endergonic by ∆G ) 26.7 kcal mol-1. The σ-(CsH) adduct of methane is less stable than separated CpRh(PMe3) and CH4 by 5.2 kcal mol-1 on the Gibbs free energy surface but more stable by 5.2 kcal mol-1 on the enthalpy surface. The difference of ca. 10 kcal mol-1 is mostly due to the loss of translational entropy upon formation of the adduct. Stronger coordination of benzene (∆H ) -15.7 kcal mol-1) leads to an η2-adduct more stable than separated reactants by ∆G ) -2.6 kcal mol-1. This η2-(CdC) adduct evolves to a σ-(CsH) complex of benzene less stable by ∆G ) 5.7 kcal mol-1 through an activation barrier of ∆Gq ) 6.7 kcal mol-1. Thus, the σ-(CsH) adduct of benzene is 2 kcal mol-1 more stable than the σ-(CsH) adduct of methane on the Gibbs free energy surface. This introduces a kinetic bias toward aromatic CsH activation that is reinforced in the TS for CsH cleavage by the significantly lower activation barrier, ∆Gq, of 3.8 kcal mol-1 for benzene compared with that for methane (∆Gq ) 7.4 kcal mol-1). The essential difference between the two TSs is the value of the Rh · · · C distance (2.23 Å, methane; 2.12 Å, benzene) leading to a late TS in the case of benzene. The stronger RhsPh bond relative to the RhsCH3 bond contributes to lower the transition state for the addition to benzene. The same argument accounts for the larger exergonicity of the reaction with ∆G ) -7 and ∆G ) -9.9 kcal mol-1 for methane and benzene, respectively. The energy diagram in Figure 4 schematically illustrates the energetics associated with the competitive reaction of methane and benzene with CpRh(PMe3)(H)2. Only the TS for CsH cleavage and the oxidative addition products are shown. It indicates that heating CpRh(PMe3)(R)(H) (R ) alkyl) in benzene solution results in irreversible formation of the phenyl hydride product.

754 Chemical Reviews, 2010, Vol. 110, No. 2

Figure 4. Calculated Gibbs free energy (kcal mol-1) diagram of the competitive reaction of CH4 and C6H6 with CpRh(PMe3)(H)2.

Thermodynamics is not always the leading factor in deciding the outcome of the transformation. Reaction of Cp*Rh(PMe3)(H)(Ph) with 1,4-C6H4F2 yields exclusively the CsH oxidative addition product Cp*Rh(PMe3)(H)(C6H3F2) featuring an ortho-F; no CsF activation product is observed.69 B3LYP calculations by Bosque et al. on CpRh(PH3) reacting with 1,4-difluorobenzene highlighted the larger thermodynamic stability of the CsF activation product.70 However, from the η2-(CdC) coordinated fluorobenzene, the energy barrier for CsH cleavage is 9.4 kcal mol-1, while it amounts to a high 34.3 kcal mol-1 for cleaving CsF. There is thus a strong kinetic bias toward CsH activation even though the corresponding product is 2.7 kcal mol-1 higher in energy than the product of CsF activation. Moreover, coordination of the fluorobenzene to the Rh fragment also favors CsH activation since the coordination of the nonsubstituted CHdCH bond is preferred over coordination of the CHdCHF bond. This shows that CsH activation is preferred because it starts from a more stable initial adduct and it is associated with a lower energy barrier. A detailed analysis of the interactions at the transition states in the case of simpler systems suggests that the oxidative addition at CsF is disfavored because of the protective effects of the fluorine lone pairs.71 Further studies on the competition between CsH and CsF activations within the same molecular systems, which will not be discussed further, have shown that different mechanisms may be operative.72–74 A further example of interplay between coordination and activation is given by thermolysis of Cp*Ir(PMe3)(C6H11)(H) in cyclohexane, which generates Cp*Ir(PMe3), which reacts in situ with H2CdCH2 to form Cp*Ir(PMe3)(CHdCH2)(H) and Cp*Ir(PMe3)(H2CdCH2) in a 2:1 ratio.75 While Cp*Ir(PMe3)(CHdCH2)(H) is stable under the thermolytic conditions of its formation, at higher temperatures it is converted cleanly to Cp*Ir(PMe3)(H2CdCH2). B3LYP calculations on CpIr(PH3)(CH3)(H) reacting with ethylene were performed by Harvey et al.76 In agreement with experiment, the η2-ethylene complex is computed to be more stable than the vinyl hydride product by 10.6 kcal mol-1. However, calculations on a singlet spin state surface failed to explain all the experimental observations. Calculations on the triplet spin state surface and characterization of the minimum energy crossing point (MECP)77,78 show that reductive elimination from singlet CpIr(PH3)(H)(CH3) yields triplet CpIr(PH3) and CH4 directly. The weak van der Waals adduct between ethylene and the triplet 16-electron fragment is the key intermediate in the reaction. The interchange of the available ethylene CsH bonds in this adduct accounts for

Balcells et al.

the observed kinetic isotope effects. Partitioning between alkene π-complexation and CsH bond activation may also occur from this common intermediate. Two-state reactivity of methane with CpM(PH3) (M ) Co, Rh, Ir) was studied at the B3LYP level by Poli et al.79 For CpM(PH3), the triplet state is always more stable than the singlet (33.3 kcal mol-1, Co; 6.2 kcal mol-1, Rh; 8.4 kcal mol-1, Ir). The interaction of methane with the triplet state of CpM(PH3) is repulsive, while it is attractive with the singlet state. Consequently, the two surfaces cross for M ) Rh or Ir, and the MECP presents a long M · · · C distance (3.067 Å, Rh; 3.350 Å, Ir). For M ) Co, the distance is short (2.312 Å), similar to that for the σ-complex on the singlet surface (2.387 Å). Thus, from singlet CpCo(PH3)(σCH4), two pathways are possible: (i) CsH oxidative addition on the singlet surface with ∆Eq ) 7.8 kcal mol-1 and ∆E ) -1.7 kcal mol-1; (ii) dissociation of methane on the triplet surface with ∆Eq ) 0.9 kcal mol-1 and ∆E ) -24.9 kcal mol-1. This explains why CsH oxidative addition has not been observed experimentally with CpCo(PR3). For Rh, the two pathways are competitive with ∆Eq ) 8.6 kcal mol-1 for CsH oxidative addition and ∆Eq ) 4.5 kcal mol-1 for CH4 dissociation. However, the oxidative addition product is 5.4 kcal mol-1 more stable than separated CH4 and triplet CpRh(PH3). CsH oxidative addition is thus possible with Rh, but the alkyl hydride product is very reactive and may easily evolve through reductive elimination. Finally, for M ) Ir, no σ-complex could be located, and the oxidative addition product is 30.9 kcal mol-1 more stable than separated CH4 and triplet CpIr(PH3), rendering reductive elimination difficult. The preceding studies point to a general picture for CsH oxidative addition by group 9 metal complexes proceeding in three consecutive steps: (1) creation of a vacant site by photolysis or thermolysis to generate a 16-electron d8 coordinatively unsaturated fragment; (2) coordination of the alkane or the arene to the vacant site to generate an intermediate σ- or π-complex (with isomerization from π(CdC) to σ(CsH) coordination in the case of arenes); (3) cleavage of the CsH bond to yield the oxidative addition product. This sequence of events was observed experimentally on the time-resolved difference IR spectra of CpRh(CO)2 in n-pentane and n-hexane following UV excitation.80 At early times (10 ps), the monocarbonyl alkane solvate CpRh(CO)(RH) is observed, as well as the parent CpRh(CO)2 compound. There is a decrease in the absorption of the alkane-solvated intermediate on a 2.5 ns time scale with the concomitant formation of the oxidative addition product CpRh(CO)(H)(R). The ultrafast dynamics of CpRh(CO)2 in cyclohexane was studied by Asbury et al., and the results showed exclusive formation of the σ-adduct of cyclohexane.81 There was no evidence for the formation of the oxidative addition product or for Cp ring slippage. This result is in agreement with the known preference of bond activation toward primary CsH sites with Rh. The isoelectronic system Tp*Rh(CO)2 (Tp* ) HB(3,5dimethyl-pyrazolyl)3) has been shown to activate CsH bonds, and the dynamics of the process has been followed on the femtosecond time scale.80,82 Photochemically induced dissociation of CO occurs in less than 100 fs, followed by coordination of the alkane in a few picoseconds to yield the solvated intermediate (κ3-Tp*)Rh(CO)(RH). Within 200 ps, κ3 f κ2 isomerization of the Tp* ligand forms the square-

CsH Bond Activation in Transition Metal Species

planar d8 ML4 σ-complex of the alkane, (κ2-Tp*)Rh(CO)(RH). The activation energy for the isomerization is estimated to be 4.2 kcal mol-1. Disappearance of (κ2Tp*)Rh(CO)(RH) follows first-order kinetics with characteristic time τ of 110 and 290 ns for activation of pentane and cyclohexane, respectively. The alkyl hydride κ2intermediate (κ2-Tp*)Rh(CO)(R)(H) evolves rapidly to the product (κ3-Tp*)Rh(CO)(R)(H). Similar results are obtained for hexane vs cyclohexane showing a faster formation of the product in the linear alkane compared with the cyclic counterpart. From the values of τ, the activation energy for CsH cleavage is between 6 and 7.8 kcal mol-1 for primary alkanes and amounts to 8.4 kcal mol-1 for cyclic hydrocarbons. DFT calculations by Zaric´ and Hall located two intermediates for CH4 reacting with TpRh(CO) (Tp ) HB(C3H3N2)3).83 A very weak η1-CH4 adduct (Rh · · · C ) 3.538 Å, RhsHsC ) 177.6°, ∆E ) -1.7 kcal mol-1) of (κ3-Tp)Rh(CO) was identified as the first intermediate seen experimentally. Dissociation of one Tp arm allowed for stronger η2coordination of methane (Rh · · · C ) 2.45 Å, RhsHsC ) 104.6°, ∆E ) -2.4 kcal mol-1). This complex was identified as the second intermediate seen after 200 ps. The calculated difference in ν(CO) vibrational frequencies for the two intermediates, 22 cm-1, is in good agreement with the value observed experimentally, 18 cm-1. Recent time-resolved infrared (TRIR) experiments combined with DFT calculations by George et al. on (Tp′)Rh(CO)2(Tp′ ) HB(4-tBu-3,5-Me-pyrazolyl)3) reacting with n-heptane, n-decane, and cyclohexane afforded new insight into the mechanism for CsH oxidative addition.84 Experimentally, species (κ3-Tp′)Rh(CO)(RH) and (κ2Tp′)Rh(CO)(RH) are shown to be in equilibrium and concentrations of both decrease simultaneously upon formation of (κ3-Tp′)Rh(CO)(R)(H). B3LYP calculations on (Tp*)Rh(CO) reacting with C2H6 showed that (κ3Tp*)Rh(CO)(RH) is an η1-adduct of C2H6 with the fragment (κ3-Tp*)Rh(CO) in a triplet spin state. The corresponding η1-adduct of C2H6 with the metal fragment in the singlet spin state lies higher in energy and evolves easily (∆Hq ) 0.9 kcal mol-1) into the η2 (CsH)-adduct with (κ2-Tp*)Rh(CO) featuring a decoordinated pyrazolyl ring. The discrepancy between the experimental (4.2 kcal mol-1) and calculated (0.9 kcal mol-1) values for the transformation (κ3Tp*)Rh(CO)(RH) f (κ2-Tp*)Rh(CO)(RH) could be lifted by considering the MECP between the singlet and triplet spin-state surfaces for the η1-adduct of C2H6. Acknowledging the known tendency of B3LYP to overestimate the stability of high spin state, a value close to the experimental one is obtained. Calculations also showed that CsH oxidative addition in (κ2-Tp*)Rh(CO)(RH) to generate (κ2Tp*)(Rh)(CO)(H)(C2H5) is not on the reaction pathway for this CsH activation process (∆Hq ) 13 kcal mol-1). The ML4 square planar complex (κ2-Tp*)Rh(CO)(RH) is very close in energy (0.3 kcal mol-1) to a distorted ML5 intermediate best described as a square-based pyramid with a long apical Rh · · · N bond (2.90 Å). This ML5 complex features a κ2.5-coordination of the Tp* ligand and leads to the lowest TS for CsH cleavage with an energy barrier ∆Hq ) 8.4 kcal mol-1 in good agreement with experiment. Because (κ2.5-Tp*)Rh(CO)(RH) both lies on the pathway between (κ3-Tp*)Rh(CO)(RH) and (κ2-Tp*)Rh(CO)(RH) and is the active species in CsH cleavage, it explains the concomitant decrease in (κ3-Tp*)Rh(CO)(RH) and (κ2Tp*)Rh(CO)(RH) upon formation of (κ3-Tp*)Rh(CO)(R)(H).

Chemical Reviews, 2010, Vol. 110, No. 2 755

Figure 5. Energy (kcal mol-1) diagram of the reaction of propane with TpRh(CNMe) showing formation of the activation product at the primary carbon (nPr-H) in preference to secondary carbon (iPr-H).

Interestingly, the X-ray structure for (Tp′)Rh(CO)2 has been determined and features this intermediate κ2.5-bonding mode of Tp.84 The computed ν(CO) frequency of (κ2.5Tp*)Rh(CO)(C2H6) is nearly identical to the one obtained for the dicarbonyl starting material, thus rendering experimental observation of (κ2.5-Tp*)Rh(CO)(RH) difficult. A similar κ2.5-coordination mode of Tp had been proposed independently by Clot et al. slightly earlier in a DFT study of the dynamic properties of propane coordinated to TpRh(CNMe).85 This followed the experimental studies by Jones of the CsH oxidative addition of alkanes and arenes to (Tp*)Rh(CNCH2CMe3) generated in situ.86,87 Reaction with propane yielded exclusively activation at the primary position, and independent synthesis of (Tp*)Rh(CNCH2CMe3)(iPr)(H) resulted in conversion to (Tp*)Rh(CNCH2CMe3)(Pr)(H) without dissociation of the alkanes as confirmed by labeling studies. The B3PW91 calculations showed that at long distance the alkane interacts with Rh in an η1-geometry and that for Rh · · · C between 3.0 and 3.5 Å one arm of the Tp ligand starts to dissociate to allow η2coordination of the alkane in a square-planar geometry. The Rh · · · N distance of ca. 3 Å and the geometry of the pyrazolyl ring are similar to those observed later by Hall. Two σ-complexes, σCH1 and σCH2, corresponding to coordination of the primary and secondary positions, are located, and the latter is less stable by 1.6 kcal mol-1. Both complexes feature coordination of a σ(CsH) bond with H trans to the long Rh · · · N bond. A transition state geometry, TS12, where Rh interacts with primary and secondary CsH bonds lies 8.4 kcal mol-1 above σCH1 and is associated with the migration from one σ-complex to the other. From σCH1, the TS for oxidative addition, OA1, is 6.4 kcal mol-1 higher in energy, whereas the energy barrier at the secondary position, through OA2, is 8.9 kcal mol-1 higher than σCH2. The oxidative addition product at the secondary position, iPrH, is 4.1 kcal mol-1 less stable than that at the primary position, nPr-H. The energy diagram of Figure 5 summarizes the results. When propane reacts with TpRh(CNMe), the two σ-complexes are likely to form. However, CsH oxidative addition in σCH2 is less facile than that in σCH1, and the energy barrier for isomerization from σCH2 to σCH1 is lower than oxidative addition at σCH2. Moreover, TS12 lies at lower energy than separated TpRh(CNMe) and propane thus explaining the labeling studies. This type of chainwalking mechanism may be general and could explain selective activation at the primary position independently of the initial interaction of the alkane CsH bonds with the metal. However, chain walking is influenced by the nature

756 Chemical Reviews, 2010, Vol. 110, No. 2

Balcells et al.

Figure 6. Sequence of transformations in the reaction of TpPt(Me)2(H) with benzene. The enthalpy (kcal mol-1) of the intermediates and TSs (above the arrow) is given relative to the separated reactants.

of the spectator ligands and is thus disfavored in presence of poorly electron-donating ligand such as trifluoroacetate.88 In the related Pt-complex TpPtMe(H)2, Keinan observed H/D scrambling of the methyl group but no methane loss when the complex is heated in CD3OD at 55 °C.89 The mPW1k calculations by Martin yielded a sequence of events similar to that shown in Figure 5. Formation of the σ-complex TpPt(H)(η2-CH4) from TpPt(H) and CH4 proceeds through a TS featuring an η2-(CsH) interaction (Pt · · · C ) 3.16 Å, Pt · · · H ) 2.288 Å) lying at ∆Gq ) 5.9 kcal mol-1 above separated reactants. The σ-complex is 6 kcal mol-1 more stable than separated reactants on the Gibbs free energy surface. It features the characteristic intermediate coordination mode of the Tp ligand already observed in the other studies with a long Pt · · · N contact (3.067 Å) trans to the coordinated H atom. From this intermediate, two pathways are described: scrambling of the coordinated hydrogen with ∆Gq ) 3 kcal mol-1 and CsH oxidative addition with ∆Gq ) 6.6 kcal mol-1 and ∆G ) -19.1 kcal mol-1. When the principle of microreversibility is applied, the σ-methane complex formed from CsH reductive coupling in TpPt(Me)(H)2 has a higher activation barrier to surmount to dissociate than to revert to the Pt(IV) complex (∆∆Gq ) 5.3 kcal mol-1). The reactivity of Tp complexes is thus strongly influenced by the ability of the ligand to change coordination mode. Jensen et al. reported the reductive elimination of methane and oxidative addition of C6D6 to form Tp*Pt(C6D5)2(D) from Tp*Pt(Me)2(H).90 From kinetic studies, the values of the activation energies for methane reductive elimination (35 kcal mol-1) and H/D scrambling (26 kcal mol-1) allowed estimation of the stability of the σ-methane complex as 9 kcal mol-1. Vastine et al. computed the reaction mechanism of the transformation TpPt(Me)2(H) + 2C6H6 f TpPt(Ph)2(H) + 2CH4 with different density functionals and basis sets.91 The B3LYP enthalpy profile is shown in Figure 6 and features the characteristic steps already encountered in CsH activation with related Tp complexes. The calculated ∆Hq of 33.8 kcal mol-1 for methane dissociation compares well with the experimental value of the activation energy and is thus identified as the rate-determining step. The TS for CsH reductive coupling computed at ∆Hq ) 24.3 kcal mol-1 is compared with the experimental activation parameters for H/D scrambling since site exchange within the σ-complex is usually considered easier than CsH oxidative addition. The overall energy profile clearly illustrates the thermodynamic driving force associated with the formation of the strong Pt-Ph bond at the expense of the weaker Pt-Me one. For benzene, η2-(CdC) coordination is preferred over η2(CsH), and the σ-(C6H5-H) complex is only marginally more stable than the corresponding TS for CsH oxidative addition. Finally, substitution of Me for Ph lowers both the

Figure 7. CsH activation by diamine and diimine cationic Pt complexes.

energy barrier for CsH reductive coupling (23.3 vs 24.3 kcal mol-1) and the energy barrier for methane dissociation (30.6 vs 33.8 kcal mol-1). In another example, reaction of (tmeda)Pt(Me)2 with [H(NC5F5)n][BArf] in NC5F5 at 0 °C gives [(tmeda)Pt(Me)(NC5F5)][BArf] (tmeda ) N, N, N′, N′-tetramethylethylenediamine, BArf)B(3,S-C6H3 (CF3)2)2). The latter complex is converted at 85 °C in the presence of benzene to the phenyl analog, while under 30 atm of 13CH4, the complex [(tmeda)Pt(13CH3)(NC5F5)[BArf] is formed.92 Related to the work of Holtcamp et al., the controlled protonolysis of (NfNf)Pt(Me)2 (Nf-Nf ) ArNdCMesCMedNAr, Ar ) 3,5(CF3)C6H3) with HBF4 · Et2O in wet CH2Cl2 gives [(NfNf)Pt(Me)(H2O)][BF4].93 The aqua complex, when dissolved in CF3CH2OH (TFE), effects the activation of methane and benzene CsH bonds under very mild conditions.93,94 These reactions can be described as the result of CsH activation by the unsaturated d8 fragment (N-N)Pt(Me)+ (Figure 7). Calculations by Heiberg et al. on the reaction of (NH2CH2CH2NH2)Pt(Me)+ with CH4 at the BP86 level located a σ-complex with an energy of -21.9 kcal mol-1 relative to separated reactants.95 The oxidative addition from this σ-complex is very easy (∆Eq ) 3.6 kcal mol-1) but not favored thermodynamically (∆E ) 1.6 kcal mol-1). In an attempt to model the influence of the solvent NC5F5, calculations have been carried out with NF2H. The latter coordinates only weakly to the σ-complex and to the TS for CsH activation but occupies the vacant site in the product of CsH activation. The significant stabilization of the latter makes the reaction exothermic by 14.8 kcal mol-1 from the σ-complex. The role of NF2H is here similar to that of the arm of the Tp ligand: partial or total decoordination allows the formation of a stable σ-complex and after CsH cleavage, recoordination provides stabilization to the product of oxidative addition. The reaction between (Nf-Nf)Pt(Me)(H2O) and CH4 was investigated by DFT calculations using a model system with HNdCHsCHdNH (N-N) as the diimine ligand.93 Coordination of methane to (N-N)Pt(Me)+ is downhill by 23.7 kcal mol-1, but water or TFE coordination is preferred (37.8 and 35.4 kcal mol-1, respectively). However, the coordination of methane to Pt with H2O or TFE interacting with the diimine backbone is only slightly uphill from the H2O and TFE complexes (6.9 and 5.3 kcal mol-1, respectively). CsH oxidative addition from (N-N)Pt(Me)(η2-CH4)+ is computed with an energy barrier of 7.9 kcal mol-1, and the product is

CsH Bond Activation in Transition Metal Species

Chemical Reviews, 2010, Vol. 110, No. 2 757

Figure 9. CsH activation by Ir(III) complex. Figure 8. CsH activation by neutral Ir(I)-methyl complex.

5.5 kcal mol-1 less stable than the σ-complex. The energy barrier is reduced to 6.2 kcal mol-1 when the σ-complex with water interacting with the diimine is considered. Interestingly, at the TS, the water has shifted from the diimine ligand to the Pt center with a Pt · · · O distance of 2.84 Å, a value similar to the Pt · · · N contacts observed in CsH activation by Tp complexes. Coordination of water to Pt while passing through the TS stabilizes the product of CsH activation, and the reaction is exothermic by 8.1 kcal mol-1. The reaction shown in Figure 7 could formally be the result of a σ-bond metathesis process (see section 2.2). Gropen and co-workers have considered this possibility in their study of diamine and diimine Pt-Me complexes.93,95 In both cases, the TS for σ-bond metathesis has a planar structure with H migrating between the two carbon atoms while featuring a short Pt · · · H distance (C · · · H ) 1.62 Å, Pt · · · H ) 1.55 Å, diamine; C · · · H ) 1.56 Å, Pt · · · H ) 1.59 Å, diimine). In the case of the diamine ligand, the σ-bond metathesis TS is 5.8 kcal mol-1 above the oxidative addition TS, whereas it is 2.6 kcal mol-1 higher for the diimine case. The energy difference between the two transition states increases to 4.8 kcal mol-1 when the water, present in the experimental conditions, is introduced in the model of the reaction. In the σ-bond metathesis pathway, the extra ligand (water) has no influence and the metal-diimine bonds are maintained, while in the oxidative TS, the water is shifted toward the Pt center and effectively stabilizes the octahedral geometry of the d6 product. The need to stabilize the d6 product in an octahedral environment appears to be a key factor for the oxidative addition to d8 metal systems. It accounts for the role of solvent or additional ligand in the case of oxidative addition to a d8 ML3 complex. Calculations on the activation of methane by oxidative addition and σ-bond metathesis have been carried out for (N-N)M(Me) (M ) Pd+, Pt+, Rh, Ir).96 The pathway through oxidative addition is favored except for Pd+. In each case, the oxidative addition is preceded by the formation of a σ-complex. The nature of the preferred pathway depends on the nature of the ligand as illustrated by a computational study (B3LYP) by Li and co-workers on (N′-N′)Pt(Ph)+ (N′-N′ ) Ar′NdCMesCMedNAr′, Ar′ ) 2,6-(Me)2C6H3) reacting with benzene.97 The energy barrier for σ-bond metathesis is 14.7 kcal mol-1, while it is 20.5 kcal mol-1 for the oxidative addition. So, unlike methane CsH activation, the σ-bond metathesis pathway is predicted to take place predominantly in this arene CsH activation. Diimine pyridine ligands provide a nice platform for CsH activation processes as illustrated by the transformation shown in Figure 8 reported by Nu¨ckel and Burger.98 BP86 calculations on the actual experimental system indicate that the reaction proceeds through a PhsH oxidative addition/ MesH reductive elimination sequence of steps. No transition state for a σ-bond metathesis pathway could be located on the PES. Bending of the methyl group away from the

molecular plane has a crucial impact on the ease of the CsH activation. Deformation of the complex to reach the geometry presenting frontier orbitals adapted to CsH activation is facilitated by CF3 substituents on the pyridine ring rendering the latter more π-accepting toward Ir.

2.1.2. d 6 Configuration After the initial discovery by Bergman of intermolecular CsH activation by the d8 Cp*Ir(PMe3) fragment, more than 10 years elapsed until a related d6 fragment proved active in CsH activation. In 1993, the Ir(III) complex Cp*Ir(PMe3)(Me)(OTf) was shown to undergo mild and selective reaction with CsH bonds in arenes and alkanes (Figure 9).99 The CsH activation has been shown to proceed after dissociation of the triflate;100 thus the active species is d6 Cp*Ir(PMe3)(Me)+. Electrospray ionization tandem mass spectrometry, in conjunction with solution-phase experiments and ab initio calculations, indicated that an intermediate of the reaction in the gas phase is Cp*Ir(η2-CH2PMe2)+. This intermediate results from intramolecular CsH activation of PMe3.101 However, studies in solution on Cp*Ir(η2CH2PMe2)(OTf) showed that the cyclometalated complex is not an intermediate in the reaction shown in Figure 9, even though it reacts faster with CsH bonds than Cp*Ir(PMe3)(Me)(OTf).102 For the reaction with CH4 shown in Figure 9, there are again two possible pathways: σ-bond metathesis and oxidative addition/reductive elimination. Both pathways are also feasible, in principle, for the cyclometalated complex. B3LYP calculations by Strout et al. on CpIr(PH3)(CH3)+ reacting with methane yielded an oxidative addition TS at 11.5 kcal mol-1 above the σ-complex, the stability of which with respect to separated reactants is -1.0 kcal mol-1.103 The Ir(V) intermediate CpIr(PH3)(CH3)2(H)+ lies at 4.4 kcal mol-1 above the σ-complex and would subsequently evolve to the product upon CsH reductive elimination. Despite an intensive search, no TS for a metathesis pathway could be located on the PES. In a later study, Niu and Hall considered the reaction of methane with CpIr(PMe3)(Me)+.104 The oxidative addition/reductive elimination pathway was not significantly altered by the substitution of PMe3 for PH3 (∆Eq ) 13.0 kcal mol-1 and ∆E ) 5.2 kcal mol-1, PMe3; ∆Eq ) 11.5 kcal mol-1 and ∆E ) 4.4 kcal mol-1, PH3). The intramolecular CsH activation to form CpIr(η2-CH2PMe2)(CH3)(H)+ was shown to be more difficult (∆Eq ) 26.5 kcal mol-1 and ∆E ) 14.7 kcal mol-1). Interestingly, the energy of the separated intermediate CpIr(η2-CH2PMe2)+ and methane is higher than the Ir(V) cyclometalated intermediate (∆E ) -5.9 kcal mol-1), and methane CsH activation is easier for the cyclometalated intermediate (∆Eq ) 5.2 kcal mol-1), in agreement with the experimental observations. These calculations confirm that the cyclometalated complex is not an intermediate in the reaction. Here again, despite a careful search, no TS for a σ-bond metathesis pathway could be found for either the inter- or the intramolecular cases.

758 Chemical Reviews, 2010, Vol. 110, No. 2

Figure 10. Formation of mixed normal/abnormal bis-NHC-Ir complex by CsH activation.

A comparison between CsH activation of cyclopropane by Cp(Ir)(PH3)(CH3)+ and CpIr(PH3) was carried out by Webster and Hall at the B3LYP level.105 In the case of CpIr(PH3)(CH3)+, the mechanism is a sequence of cyclopropane oxidative addition/methane reductive elimination that leads to endo and exo cyclopropyl complexes of similar energy. These complexes transform to η3-allyl isomers after formation of an Ir(CH2)(CHdCH2) intermediate through CsC oxidative addition. The rate-determining step is the CsH cleavage of cyclopropane with ∆Eq ) 13.1 kcal mol-1. In the case of the reaction with CpIr(PH3), a σ-complex of cyclopropane with a stability of -7.5 kcal mol-1 could be located, and the TS for CsH cleavage is only 0.1 kcal mol-1 above the σ-complex. From the same σ-complex, a TS for CsC cleavage to form an iridacyclobutane could be located with ∆Eq ) 2.8 kcal mol-1. The product of CsC activation, which is the thermodynamic product, is computed to be 15.8 kcal mol-1 more stable than the hydrido-cyclopropyl product. CsH activation by Ir(III) complexes proceeds through an Ir(V) intermediate that is expected to be less stable. Thus any factor leading to stabilization of the Ir(V) intermediate should ease CsH activation. Webster and Hall have addressed the computational design of stable Ir(V) complexes.106 The reaction (ηx-L)L′IrIIIR + R′H f (ηxL)L′IrVRR′H (ηx-L ) Cp, Tp, and carborane; L′ ) PR3, CR3-, SiR3-, and SnR3-; R′ ) alkyl, CR3, H) has been studied at the DFT level. Finding a ligand set that allows “trapping” of an Ir(V) intermediate has proven difficult. Improved stability of an Ir(V) intermediate is obtained in the reaction of methane with Cp*Ir(PMe3)(H)+. When neutral complexes, obtained by substitution of PR3 with silyl or stannyl, are considered, the Ir(V) intermediate and the Ir(III) precursor have similar energies (∆E < 0 but ∆G > 0). In an attempt to form abnormal bis-N-heterocyclic carbene (NHC) complexes from C2-methylated bis-imidazolium salts, Viciano et al. observed an interesting competition between CsH activation at sp2 and sp3 carbon atoms (Figure 10).107 B3PW91 calculations indicated that the first CsH activation by Cp*IrI2 is easier at the sp3 than at the sp2 position (∆Eq ) 31.6 vs 34.1 kcal mol-1, respectively). The second CsH activation is an intramolecular process at the d6 Ir(III) complex Cp*Ir(CH2-R)(I). CsH activation of the methyl group of the second imidazolium ring has an energy barrier of 24.6 kcal mol-1 thus showing the ability of the ligands on Ir to modulate the energetics of the CsH cleavage by making the metal center more electrophilic. However, the TS for CsH activation at the sp2 position of the second imidazolium to give the observed product is favored by 7.1 kcal mol-1 as a result of prior coordination of the C4dC5 π bond. The calculations have thus highlighted the sequence responsible for formation of the mixed ligand: intermolecular sp3 CsH activation first, followed by intramolecular sp2 CsH activation. The role of the base AcO- is to deprotonate the Ir center after each CsH oxidative addition step, thus providing a thermodynamic driving force for the reaction. A comparative study of methane CsH activation by TpRe(CO)2 and CpRe(CO)2 was carried out by Bergman et al. at the B3LYP level in order to understand why oxidative

Balcells et al.

Figure 11. Systems computed in the study by Su and Chu.122

addition is observed with CpRe and not with TpRe.108 The σ-complexes were computed to have similar stability with respect to separated reactants (-6.3 kcal mol-1, Tp; -8.7 kcal mol-1, Cp), but the TS for CsH activation with CpRe(CO)2 is 9.1 kcal mol-1 more stable than the TS for CsH activation with TpRe(CO)2. Consequently the reaction is exothermic for CpRe(CO)2 (-7.9 kcal mol-1) and endothermic for TpRe(CO)2 (+6.4 kcal mol-1), in good agreement with the experimental observations. The different behavior is ascribed to the steric influence of the Tp ligand that disfavors the seven-coordinated geometry of the oxidative addition product. Clot et al. studied the mechanism of arene CsH activation by CpRe(CO)2 at the B3PW91 level.109 As expected, the π-benzene adduct is more stable than the σ-adduct (4.3 kcal mol-1), and the latter is a precursor for the CsH cleavage step, which gives a cis isomer (∆Eq ) 9.1 kcal mol-1 and ∆E ) 6 kcal mol-1). This cis isomer subsequently evolves exothermically (∆E ) -4.9 kcal mol-1) to the trans isomer with an energy barrier of 24 kcal mol-1. The rate-determining step in this transformation is not the CsH activation but an isomerization process of the oxidative addition product. The trans isomer is computed to be less stable than the π-benzene adduct by 5.3 kcal mol-1, in agreement with the experimental observations where no activation is seen for benzene.110 Legzdins et al. have studied the intermolecular activation of hydrocarbon CsH bonds by Cp*W(NO)(PMe3).111 The role of fluoro substituents on the CsH activation in arenes will be further discussed in section 3.1.

2.1.3. d 4 Configuration The d4 fragment Cp2W, which can be formed by solution photolysis of Cp2W(H)2 or Cp2W(CO), has been shown to activate CsH bonds.112–116 NMR studies of methane elimination from Cp2W(Me)(H) provided insight into the process of reductive elimination and hence of its microscopic reverse oxidative addition.117,118 Other metallocene complexes have been studied (Cp*2W(Me)(H)119 and Me2CCp2W(Me)(H)120), and they all show significant differences in their thermal stability. The non-ansa methyl-hydride complexes decompose over the range 40-100 °C, producing methane, while the ansa-bridged compound is thermally stable up to 120 °C. Several computational studies have addressed the mechanism of methane elimination from group 6 metallocene methyl-hydride complexes.121–125 Su and Chu carried out B3LYP calculations on various metallocenes (Figure 11, M ) Mo, W).122 They computed the singlet and triplet spin states for the d4 metallocene fragments, but the oxidative addition of methane was considered only on the singlet surface. The oxidative addition is composed of two steps: methane coordination to form a σ-complex and CsH oxidative cleavage. The latter step is computed to be more facile for the structure with parallel Cp rings than for the bent geometry of the ansa system. This is due to the larger stability of the σ-complex with the ansa-metallocene based on the shapes of its frontier orbitals. The singlet-triplet gap, which is smaller for Cp2W than ansa-Cp2W (5.2 kcal mol-1

CsH Bond Activation in Transition Metal Species

Chemical Reviews, 2010, Vol. 110, No. 2 759

Figure 12. Schematic representation of the mechanism for methane oxidative addition to triplet group 4 metallocenes.

vs 18.9 kcal mol-1 with triplet more stable as shown experimentally by Perutz126,127), is shown to determine the relative rate of reactions, following a rationale in terms of configuration mixing.128 DFT calculations were also performed by Green et al. on Cp2M, H2CCp2M, and Cp*2M (M ) Mo, W) reacting with methane.121,123 The d4 metallocenes have a triplet ground state, but the CsH coordination and cleavage occur on the singlet surface. To rationalize the chemistry of these systems, it is important to determine how the singlet and triplet surfaces cross. Therefore, the minimum energy crossing point (MECP) was determined for Cp2M (M ) Mo, W), and the PES for methane activation on the singlet surface was also calculated (Figure 12). Table 2 collects the values of the energies ∆Ei (kcal mol-1, i ) 1-3 see Figure 12). From these values, it is clear that the rate-determining step is crossing the MECP for the ansa-W complexes and the CsH cleavage for the other complexes. The energy barrier is significantly lower for the ansa-W complexes, in agreement with the expected greater reactivity. The small energy gap between the singlet and triplet H2CCp2W complex is a result of the constraint imposed by the tether and is at the origin of the greater reactivity. The triplet, which prefers the two cyclopentadienyl rings to be parallel, is destabilized in the ansa complex and is thus more reactive. Any photogenerated ansa-Cp2W fragment immediately reacts with any CsH bond present in the solution. This explains the experimental observations that an ansa-methyl hydride W complex is stable even at 120 °C. The energy barriers for reductive elimination are in goodagreementwiththeavailablevaluesforCp2W(CH3)(H)117,118 and Cp*2W(CH3)(H).119 For Cp2W(η2-CH4), the energy barrier for H exchange is lower than that for reinsertion, while it is of the same order for Cp*2W(η2-CH4), explaining the H/D scrambling observed experimentally. The preceding study was completed by a characterization of the mechanism for methane oxidative addition to H2SiCp2M and Me2SiCp*2M (M ) Mo, W) by Harvey et al.125 Here again, the relative energy of the MECP and the TS for oxidative cleavage is the crucial factor. Experimentally, it was shown by Churchill et al. that methane reductive

Figure 13. Various mechanisms for formal σ-bond metathesis.

elimination from Me2SiCp*2M(Me)(H) is characterized by an inverse primary kinetic isotope effect for M ) W but a normal one for M ) Mo.129 For M ) Mo, the CsH coupling step is rate-determining (Table 2), and thus KIE is normal. For M ) W, the situation is reversed with methane dissociation through the MECP being now the rate-determining step (Table 2). Thus the experimental KIE is an equilibrium isotope effect, and the inverse value reflects the tendency for deuterium to occupy the sites associated with high vibrational frequencies (η2-(CsH) vs MsH). However, kinetic isotope effects are not always straightforward to analyze, and Parkin has shown how transition from a normal to an inverse behavior could be obtained by changing the temperature.130,131

2.2. σ-Bond Metathesis σ-Bond metathesis is a one-step reaction by which two σ bonds are broken and two new σ bonds are formed in a concerted manner without change of the metal oxidation state (Figure 13a). A recent review describes the current understanding for this reaction.132 This mechanism has been proposed for early and late transition metal complexes. However, in the case of the late transition metal complexes, where the metal can have various oxidation states, the same transformation can be formally achieved either by σ-bond metathesis or by successive oxidative addition and reductive elimination (Figure 13b). Preference for either of these mechanisms has been the topic of a number of theoretical studies, which will be presented in section 2.4. In this section, we focus on the case of early d0 transition metal complexes for which a change in the oxidation state of the metal is not energetically accessible. Activation of CsH bond by σ-bond metathesis is prevalent in the case of groups 3 and 4 metal complexes, as well as with lanthanide and actinide complexes. Hydrogen transfer between ligands (Figure 13c), which involves a transition state analogous to that of classical σ-bond metathesis, is also achieved without modification of the oxidation state at the metal, and some representative cases will be presented. Computational studies have focused on metallocene complexes, which have been found to be very useful as catalysts

Table 2. Energies (kcal mol-1) in the Process of Methane CsH Activation by Various d4 Metallocenesa ∆E1 ∆E2 ∆E3 a

Cp2Mo

Cp2W

Cp*2W

H2CCp2W

Me2SiCp*2Mo

Me2SiCp*2W

21.8 28.2 9.8

17.8 18.0 -11.0

25.0 25.8 -24.4

6.5 3.4 -29.2

11.1 15.4 -4.0

11.8 9.4 -21.6

See Figure 12 for definition of energies.

760 Chemical Reviews, 2010, Vol. 110, No. 2

for olefin polymerization and have considerable potential for aliphatic and alicyclic CsH bond activation.133,134 The mechanistic aspects of the CsH activation have also been studied in relation to the Ziegler-Natta polymerization process notably by Bercaw and co-workers.135,136 We will not review the computational works related to polymerization. However, we cite a key computational study on CsH activation in relation to chain termination for the interested reader.137 One of the most emblematic reactions involving CsH activation with early transition metal complexes has been the activation of methane by (C5Me5)2LuCH3 for which the degenerate reaction of exchange of methyl groups has been proven by isotope labeling.138,139 The CsH bond activation is also a key step in the reaction of CH3X with metallocene cerium complexes.140,141 The competition between CsH and CsX bond activation in fluorarenes by the same cerium complexes has been studied.142,143 Competition between CsH and CsX (X ) Cl) has been observed in Ti(IV) complexes, and the mechanism of this reaction has also attracted the interest of computational chemists.144,145 The catalytic hydromethylation of olefins by a metallocene of scandium is an example of methane functionalization.146,147 The key structural and electronic features of the transition state for σ-bond metathesis were established in the study of the H for H exchange reaction between a metal hydride complex and H2. The earlier studies, which included a generalized valence bond (GVB) analysis of the reaction,148,149 were followed by DFT calculations by Ziegler150 and by Maron and Eisenstein,151 carried out for reactions with Sc, Y, and lanthanide complexes. Earlier calculations have used chloride ligand to represent the cyclopentadienyl ligand. In the more recent studies, C5H5 and C5Me5 have been used in the models, and various levels of calculations have been tested for C5Me5.152,153 For the calculations of lanthanide elements, large core pseudorelativistic effective core potentials have been used. This methodology, which puts the inner 4f electrons in the core, has been validated by comparison with calculations using small core pseudorelativistic effective core potentials.154 The transition state for H for H exchange between the MsH bond of X2MH and H2 has a kite-shaped geometry with the metal and the three hydrogen atoms lying in a plane bisecting the XsMsX angle. This applies to any model used to represent the metallocene (X ) Cl or Cp) and for any group 3 or lanthanide metal. In principle, the reaction starts by the coordination of the substrate to the electron-deficient metal center. However, the stability of the adduct relative to separated components is weak due in particular to the absence of back-donation from the metal to the empty orbitals of the substrate. Its stability, in the case of H2 and Cp2ScH, is less than 4 kcal mol-1.150 No equivalent adduct was found in the case of Cp2LnH.151 Adducts of CH4 are also found to be weak even though the bonding energy of CH4 depends on the model used for the metal fragment. The presence of a stable adduct between the electron-deficient complex and the reagent (H2, CH4) is thus not an important feature in the σ-bond metathesis reaction in the case of d0 metal fragment and will not be discussed further. The geometrical features of the transition state deserve further comments. In a study of the reaction of Cp2ScH with H2, Ziegler et al. mention the presence of a short contact between the metal and the β-H.150 They also state that the ScsH bond lengths at the transition state for the H for H

Balcells et al.

Figure 14. Transition state for σ-bond metathesis.151

Figure 15. Different regiochemistry in σ-bond metathesis.

exchange are only slightly longer than a normal ScsH bond length. These facts account for the low energy barrier. In a study of the reaction of Cp2LnH with H2, Maron and Eisenstein obtained similar geometrical results.151 The energy barriers are low and similar for all Ln(III) metals. The LnsH bonds at the transition state are just slightly longer than the LnsH bond in Cp2LnH. The authors also note that the HsHsH angle is around 156° for any Ln(III) element, which leads to a shorter MsH distance for the β- than for the R-hydrogen (Figure 14). For instance, in the case of Cp2LaH, the La-H distance for the R-hydrogen is 2.276 Å, while it is 2.262 Å for the β-hydrogen and it is 2.142 Å in Cp2LaH. The large HsHsH angle at the transition state suggests that the three hydrogen atoms have accumulated a negative charge, localized on the hydrogens near the metal, and take the geometry of a H3- moiety, in interaction with the metal fragment; this fact is supported by a charge analysis.155 Therefore, these authors consider that the reaction is best viewed as a proton transfer between two hydrides in the stabilizing field of a positively charged metal center. The positive charges at both the metal and the migrating H exclude any interaction between them, despite the short internuclear M · · · H distance. The transition state for σ-bond metathesis between the MsC bond of Cp2MsR and the CsH bond of R′sH has similar geometrical features as described above: the fourmembered ring has a kite-shaped geometry and the metal, the carbon of R bonded to M, and the activated CsH bond lie in a plane, bisecting the CpsMsCp angle. Two transition states, corresponding to two different reactions, are in principle possible. One reaction corresponds to an R for R′ exchange, and the hydrogen atom occupies the β-position at the transition state (Figure 15a). The other reaction forms RsR′ and MsH, and R′ occupies the β-position at the transition state (Figure 15b). These two reactions are found to have dramatically different energy barriers in the case where R′ is an alkyl group. Only the reaction with the hydrogen atom at the β-position at the transition state (Figure 15a) is calculated to have accessible energy barriers. If R′ ) H, the reaction is calculated to be exothermic for the formation of the metal hydride. In other words, the metal hydride does not activate the CsH bond of an alkane, while the metal-alkyl bond is hydrogenolyzed by H2 in agreement with experiment. These results have been found for either Cp2ScH150 and Cp2LnH complexes.156–158 The energy barrier of the reaction shown in Figure 15a varies between 5 and 19 kcal mol-1 depending

CsH Bond Activation in Transition Metal Species

Figure 16. Charge distribution at the transition state for σ-bond metathesis.

on the model used for the ligand (lower for Cl than for Cp) and the nature of the metal. In contrast, the reaction of Figure 15b has an energy barrier in the range of 37.5 kcal mol-1 relative to the adduct between Cl2ScCH3 and CH4.159 For the reaction of H for H exchange between Cp2LnH and CH4 (CH3 at the β position), the activation barrier has been calculated to be around 70 kcal mol-1 for any Ln(III) complexes.157 The rationale for the very different barriers of these two reactions has been discussed notably by Maron and Eisenstein. Following the analysis for the reaction of Cp2MH with H2, the transition state for the reaction of Figure 15a can be viewed as a proton transfer between two negatively charged R and R′ alkyl groups (Figure 16). This is supported by the structure of the transition state: for instance, in the reaction of Cp2LaCH3 with CH4, the CsHsC angle is equal to 180°. The charge distribution is similar when the alkyl, such as Me, is at the β-position. The positively charged alkyl group is thus interacting with two negatively charged groups (H and R). The result is that the [H(CH3)R]- ensemble has an electronic structure similar to that of CH5-, which is not a favorable situation. To support this interpretation, calculations show that replacing C by Si at the β-position yields considerably lower barriers.160 This is in agreement with the commonly accepted idea of the ability of Si to become hypervalent as in SiH5-. The accessibility of a σ-bond metathesis reaction with a silyl group and not with the methyl group at the β-position is central to the understanding of the reactions between alkyl and aryl silanes with Cp*2Sm derivatives.161,162 The discrimination between paths with H or alkyl at the β-position has been used to rationalize the H for D exchange on Zr-H supported by silica.163 It has been shown by experiment that the rate of the reaction of Cp*2M-CH3 with CH4 decreases in the order Y > Lu > Sc. This comparison was studied with DFT calculations by Cramer164 and by Maron and Eisenstein.158

Chemical Reviews, 2010, Vol. 110, No. 2 761

The two groups of authors have compared the influence of Cp vs Cp* on the reactivity trends and have found that introducing Cp* is necessary for reproducing the experimental observations. Maron and Eisenstein mention that it is necessary to consider the charges developed on the four centers at the transition state in order to understand the reactivity better. Cramer et al. have also studied the reactions that take place in competition with the bimolecular methyl exchange (Figure 17). They have shown that the unimolecular methane ejection via formation of a tuck-in complex competes with the bimolecular methyl exchange reaction for Sc but not for Y and Lu. The greater propensity for Sc to form the tuck-in complex is related to the shorter metal-ligand distances observed for this metal atom. The calculated enthalpy for CsH activation of methane is 24.1 kcal mol-1 via the bimolecular process, which is significantly higher than the experimental value of 11.6 kcal mol-1. The enthalpy of activation is lowered to 19.2 kcal mol-1 by including the tunneling contribution. This latter study was extended to the reaction of methane with Cp*2MR (M ) Sc, Lu; R ) Me, CH2C(CH3)3), with the aim of understanding the influence of the R group on the metathesis reaction.165 With the mPW1K density functional, the bimolecular reaction is calculated to dominate but the unimolecular reaction, proceeding through the tuck-in complex, becomes increasingly competitive for bulky alkyl groups and for metals with small ionic radius. The quantummechanical tunneling is predicted to increase the overall reaction rates by 1-3 orders of magnitude over the temperature range 284-323 K. This study was repeated for ansa(bis(η5-2-indenyl)methane)MR complexes (M ) Sc, Y, Lu; R ) CH3,CH2C(CH3)3).166 It was shown that the presence of an ansa bridge disfavors the unimolecular reaction via the tuck-in complex and favors the bimolecular reaction. The full path for hydromethylation of propene, where several σ-bond metathesis processes are in competition, was studied for metallocene complexes of scandium and lutetium (Figure 18).167,168 The formation of secondary products, notably resulting from various CsH activations, was studied in addition to the main pathway. Vinylic and allylic CsH activation of propene are competitive with the main path and are responsible for the formation of the secondary products. These reactions, although releasing isobutane, destroy the catalyst since no regeneration from CsH activation of methane is possible. The elementary steps responsible for

Figure 17. Competition between intramolecular (top, unimolecular methane dissociation) and intermolecular (bottom, bimolecular methyl group exchange) σ-bond metathesis reactions.

762 Chemical Reviews, 2010, Vol. 110, No. 2

Balcells et al.

Figure 18. Postulated mechanism for the hydromethylation of propene.

Figure 20. Products of the reaction of Cp*(tBu3PdN)TiH+ and halobenzene.

Figure 19. Reactions between Cp2CeH and arenes: (a) transition state that could not be located; (b) pathway for formation of Cp2Ce(C6F5) from Cp2CeH and C6F5H.

the disappearance of the catalyst have lower energy barriers in the case of lutetium, which excludes the Lu complex as an efficient catalyst. A similar study was carried out for the reaction of propene and isobutene with ansa-H2Si(C5H4)2ScCH3, a model for ansaMe2Si(C5Me4)2ScCH3.168 It has been concluded that the hydromethylation is more efficient for isobutene than for propene, in agreement with experiment. In particular, the vinylic CsH activation has an accessible energy barrier and produces intermediates like Me2Si(C5Me4)2Sc(CHd(CH3)2). Monomeric Cp′2CeH (Cp′ ) 1,2,3-(Me3C)3(C5H2)) reacts with C6F5H to form Cp′2CeF and H2, via an intermediate, Cp′2Ce(C6F5), which was isolated.142 DFT(B3PW91) calculations of the reaction of C6F5H with Cp2CeH to form Cp2Ce(C6F5) and H2 were carried out. A transition state with the phenyl group at the β-position was searched for (Figure 19a). Despite considerable effort, no such transition state could be located. The proposed pathway starts by the coordination of C6H5F to the empty coordination site of Cp2CeH by way of the fluorine ortho to the CsH bond (Figure 19b). In this adduct, the arene lies in the plane bisecting the Cp2Ce angle. This geometry brings the positively charged H of C6F5H close to the hydride. The activation barrier to form Cp2Ce(C6F5), which is an observed intermediate, and H2 from the adduct Cp2Ce(C6F5H)H is 10.2 kcal mol-1. The results obtained in this study show that the reactivity pattern found for alkanes is also applicable to arenes. Thus, a transition state of σ-bond metathesis with a carbon at the β-site is in general energetically unfavorable. The reaction of Cp*(tBu3PdN)TiH+ and C6H5X (X ) H, Cl, Br) leads to different products depending on the pressure of H2 used in the experiment (Figure 20).144 At low pressure of H2, the release of H2, through ortho-CsH activation and the formation of the β-halophenyl cation Cp*(tBu3PdN)Ti(2X-C6H4)+ has been observed. This halophenyl complex evolves further. At higher pressure of H2, an H for X exchange is observed with release of benzene. Calculations were carried out with DFT(TPSS) method.145 The pathway

Figure 21. Cp′2CeH and associated metallacycle.

for the ortho-metalation of the halobenzene is found to be similar to that described for Cp2LaH and C6F5H.142 However, for the H for X exchange, a direct σ-bond metathesis pathway with a kite-shaped transition state has been located with an enthalpy of activation of 20.1 kcal mol-1 above the halophenyl complex (X ) Cl). The authors also mention that using B3LYP and B3PW91 methods, they also located a kiteshaped transition state for the H for F exchange between Cp2LaH and C6F6 (∆Hq ca. 6 kcal mol-1, ∆Gq ca. 9 kcal mol-1), in contrast to the results published previously.142,169 We have described earlier the reasons why the kite-shaped transition state of σ-bond metathesis with an alkyl group at the β-position has a high energy. A way to lower the energy of the transition state is to replace C with Si. Another way is to substitute the H atoms on the alkyl group by electronwithdrawing atoms. This proposal was tested by calculating the energy profile for the H for F exchange in the reaction of Cp2LnH with CF4. Although the energy barrier is significantly lower than that associated with CH4, it is still too high for the reaction to be feasible.170 This was confirmed by Andersen by the absence of any reaction between Cp′2CeH (Cp′ ) 1,2,4-(Me3C)3(C5H2)) (Figure 21) and CF4. However, CH3F, CH2F2, and CHF3 react with Cp′2CeH. In all cases, Cp′2CeF is formed as well as CH4, but the rate decreases in the order CH3F < CH2F2 , CHF3.140 The DFT calculations, carried out with Cp2CeH as a model for Cp′2CeH, show that the σ-bond metathesis has high activation barriers for all fluoromethane derivatives (∆Gq ) 31.1 kcal mol-1, CH3F; ∆Gq ) 33.2 kcal mol-1, CH2F2; ∆Gq ) 41.1 kcal mol-1, CHF3). The preferred pathway is found to be a two-step pathway (Figure 22). In the case of the reaction with CH3F, the first step is a proton transfer to form Cp2Ce(CH2F) and H2. In this intermediate, the CH2F group is dihapto-bonded to the metal center. This intermediate further evolves by inserting CH2 into H2 to form CH4. As CH2 inserts into H2, the CsF bond is cleaved, and the final

CsH Bond Activation in Transition Metal Species

Chemical Reviews, 2010, Vol. 110, No. 2 763

Figure 22. Two-step pathway for the reaction of Cp2CeH with CH3X.

products Cp2CeF and CH4 are formed. The proton transfer has a lower barrier than the following step, which involves the CsF cleavage, is rate-determining. The reaction pathways are calculated to be similar for the CH2F2 and CHF3 substrates, which yield the Cp2CeCHF2 and Cp2CeCF3 intermediates, respectively. While the activation barrier of the proton transfer step does not vary much with the number of fluorines and stays low in all cases (∆Gq < 18 kcal mol-1), the activation barrier of the transfer of the carbene into H2, which increases in the order CH2 < CHF < CF2, determines the overall rate of the reaction. Repetition of the sequence allows defluorination of the haloalkane. Experimental support for this two-step mechanism comes from the characterization of several products resulting from the insertion of CH2 into CsH bonds, as well as addition of CH2 into various traps. This two-step pathway, which is best viewed as a carbenoid pathway, was generalized through a combined experimental and theoretical study of the H for X exchange in the reaction of CH3X (X ) F, Cl, Br, I, OMe, and NMe2) with Cp′2CeH and the associated metallacycle (Figure 22).141 It was shown that the reactions fall into three distinct classes: (i) class a (X ) Cl, Br, and I) rapidly forms Cp′2CeX and CH4 without formation of identifiable intermediates in the 1 H NMR spectra, (ii) class b (X ) OMe) proceeds rapidly to Cp′2Ce(η2-CH2OMe) and H2, which is sufficiently stable to be isolated and characterized, and then to Cp′2CeOMe and CH4, and (iii) class c (X ) NMe2) does not result in formation of Cp′2CeNMe2, but deuterium labeling experiments show that H for D exchange occurs in NMe3. The DFT calculations show that H for X exchange via one-step synchronous σ-bond metathesis is associated with the highest activation barrier in all cases. The preferred pathway is a two-step pathway, in which the first step, with a low activation barrier, is an R-CsH activation, i.e., a proton transfer, that forms Cp2Ce(η2-CH2X) and H2, followed by trapping of CH2 by H2, which proceeds with a higher activation barrier, its height defining the classification as either a, b, or c. The free energy profiles show that the free energy barrier of the proton transfer is similar for all X (between 13.9 kcal mol-1 for X ) OMe and 18 kcal mol-1 for X ) I). However, the values do not follow the experimental gas-phase proton dissociation enthalpies for CH3X. Thus, the activation barrier is not just determined by the acidity of the R-CsH bond, even though the relative acidity is the primary reason for the higher activation energy when H2 and CH4 are compared,171 and the identity of X plays a role in determining the barrier. While these reactions are formally H for X exchange, they do not occur by means of the one-step σ-bond metathesis but by a two-step mechanism initiated by a CsH activation process. At the origin of this preference are the low barrier associated with the CsH bond activation and the high barrier associated with a σ-bond metathesis transition state with a carbon at the β-position. A very recent study of the amine-catalyzed benzene CsH bond activation in Cp*2ZrMe+ olefin polymerization catalysts

Figure 23. Hypothetical catalytic cycle for hydrocarbon heterofunctionalization through CsH 1,2-addition across MdX.

Figure 24. CsH activation by (silox)2(tBu3SiNH)TiR through 1,2addition.

has highlighted a new pathway for the deactivation route in aromatic solvent.172 The external base NMe3 deprotonates the coordinated benzene molecule and protonates the methyl group, thus yielding formally a metathesis reaction. This twostep pathway was shown by DFT calculations to be preferred over the one-step direct σ-bond metathesis pathway.

2.3. 1,2 Addition 2.3.1. Early Transition Metals When a ligand with π-bonding capability (alkylidene, alkylidyne, amide, imide, alkoxide, aryloxide) is coordinated to an early transition metal, the unsaturation at the metal center is stabilized through π-donation from the ligand. The MdX functionality (X ) CR2, CR, NR2, NR, OR) adds to a CsH bond in a [2σ + 2π]-type reaction (Figure 23). The 1,2-addition product M(XH)(R) could lead to a heterofunctionalized hydrocarbon upon reductive elimination. If the regeneration of the MdX group could be made catalytic, this would constitute a major advance. A key step in such a putative catalytic cycle is net 1,2-addition of a CsH bond across a metal-oxygen or metal-nitrogen bond. The CsH cleavage in itself is only one of the several steps that need to be achieved. As in the case of oxidative addition reactions, a vacant site on the metal should be created for the CsH bond to coordinate and engage in an interaction with MdX. A combined experimental and computational study by Wolczanski, Cundari, and co-workers on (silox)2(tBu3SiNH)TiR (silox ) tBu3SiO; R ) Me, Et, CH2Ph, CHdCH2, tBu, nBu, Ph, H, cPr, cPe) illustrates the basic aspects of the 1,2-addition.173–175 Thermolysis of (silox)2(tBu3SiNH)TiR in benzene-d6 produces RH and (silox)2(tBu3SiND)Ti(C6D5), consistent with the mechanism shown in Figure 24. Activation parameters for RH elimination were determined for R ) Me and cPe. They indicate significant bond-breaking (∆Hq ca. 20 kcal mol-1) occurring within a constrained, unimolecular transition state (∆Sq ca. -12 cal K-1 mol-1), supporting a planar geometry for RsTisNsH in the TS. Large primary kinetic isotope effects were observed for 1,2-CsH elimination in agreement with a four-center TS with a relatively linear R · · · H · · · N geometry featuring similar amounts of NsH bond breaking and CsH bond-making.173 Ab-initio calculations (MP2//RHF) were performed on the model system (HO)2TidNR′ reacting with RH (R ) Me,

764 Chemical Reviews, 2010, Vol. 110, No. 2

Figure 25. CsH activation by (PNP)Ti(tCR).

Et, CHdCH2, cPr, Cy, Ph, or CH2Ph and R′ ) H; R ) Me, c Pr, or Cy and R′ ) SiH3; R ) Me and R′ ) SiMe3).175 Increasing the size of the R′ group results in a more polarized Tiδ+sNδ- bond and a more stable σ-adduct between Ti and RH. The structure and the energetics of the substrate/Tiimido interaction is considered to be the main factor that determines the selectivity between activation of the various CsH bonds in RH. Molecular mechanics calculations on the various adducts between (silox)2Ti(NSitBu3) and isopentane are in agreement with primary CsH bond preferentially coordinated versus secondary and then tertiary CsH bonds. Sakaki et al. studied in detail the reaction of (Me3SiO)2Ti(dNSiMe3) with methane.176 DFT, MP2 to MP4(SDQ), and CCSD(T) calculations were carried out. The B3LYP binding energy of methane is -4.5 kcal mol-1, and from this σ-adduct, the kinetic and thermodynamic parameters for CsH activation are ∆Eq ) 14.6 kcal mol-1 and ∆E ) -18.2 kcal mol-1. Formation of the σ-complex is associated with a diminution of the electron population on CH3 and H concomitant with an increase of the Ti electron population, indicative of a ligand to metal charge transfer (electrophilic attack). The passage through the transition state is accompanied by a significant increase of electron population on CH3 and a significant decrease of electron population on H, which suggests that the CsH cleavage is heterolytic. These population changes are due to electron transfers from populated σ(CsH) to vacant Ti dz2 MOs and from populated dπ-pπ(TidN) to vacant σ*(CH) MOs. An energy density analysis of the potential energy surface by Nakai et al. confirmed that Ti, N, C, and H are the active sites in the system.177 Mindiola, Baik, and co-workers studied intermolecular CsH bond activation promoted by a titanium-alkylidyne complex.178,179 The reactive TitCR functionality is generated from a titanium alkyl alkylidene through R-H migration (Figure 25). Experimentally, the R-migration is shown to be the rate-determining step. This is confirmed by DFT calculations, which show that the TS for R-migration is 27.8 kcal mol-1 above the reactant, whereas the TS for 1,2-addition of benzene is 21.0 kcal mol-1 above the same reference. Although the reactive Ti-alkylidyne intermediate is only 4.6 kcal mol-1 less stable than the reactant, its energy prevents any direct experimental observation. The overall reaction is computed to be exothermic by 11.8 kcal mol-1.178 Coordination of the CsH bond to the Ti-alkylidyne complex is not selective, as σ-adducts of CMe4 and C6H6 are computed to have similar energies. However, there is a clear distinction between C(sp3)sH and C(sp2)sH bond

Balcells et al.

cleavage in these σ-adducts. The former is 6.9 kcal mol-1 more difficult than the latter (∆Gq ) 12 kcal mol-1, CMe4; ∆Gq ) 5.1 kcal mol-1, C6H6). Charge analysis again showed a build-up of negative charge on C and a build-up of positive charge on H in the TS for 1,2-addition, in agreement with an heterolytic cleavage of the CsH bond. Concomitantly, the TitC bond polarity Tiδ+sCδ- increases in the TS. Experimentally, for the reaction of the Ti-alkylidyne complex with CsH bonds, it was shown that CsH cleavage is the rate-determining step for C(sp3)sH, whereas CsH bond coordination is rate-determining for C(sp2)sH. OsCl3(CsCHdCPh2)(PPh3)2 is reduced to the indenyl complex {(η5-C9H6(Ph)}OsCl(PPh3)2 by Zn in the presence of PPh3.180 An unusual metallanaphthalyne complex is isolated as an intermediate during this reaction. DFT calculations on the model complex OsCl(PH3)2(CsCHdCHPh) show that the transformation is initiated by an aryl CsH oxidative addition at the osmium center, even though a carbyne functionality is present and could have been involved in a 1,2 addition. Legzdins has studied in detail the thermal activation of hydrocarbon CsH bonds by Cp*M(NO)(CHR) complexes of molybdenum and tungsten, and some significant results are represented in Figure 26.181 Poli and Smith have carried out B3LYP calculations on CpM(NO)(CH2) (M ) Mo, W) reacting with CH4.182 They located σ-complexes on the PES and showed that the CsH cleavage is more difficult for Mo than for W, as shown by the higher energy barriers (∆Eq ) 14.3 kcal mol-1, Mo; ∆Eq ) 7.8 kcal mol-1, W) and the smaller exothermicity (∆E ) -24.9 kcal mol-1, Mo; ∆E ) -35.7 kcal mol-1, W). The lower stability of the dimethyl Mo complex is in agreement with the milder conditions used experimentally (20 °C, 30 h, Mo; 70 °C, 40 h, W) to generate the active alkylidene complex (Figure 26). A combined experimental and theoretical study of the reaction mechanism was carried out by Legzdins et al. on the tungsten complex.183 The KIE values were inconsistent with rate-determining CsH bond activation by the MdC linkage and indicated that coordination of the substrate to the metal center is the discriminating factor in alkane and arene intermolecular competitions. B3LYP calculations on CpW(NO)(CH2) reacting with toluene probed this competition through characterization of the extrema along C(sp3)sH and C(sp2)sH activation. The π-arene complexes (two orientations with respect to Cp) are computed to be ca. 10 kcal mol-1 more stable than the σ(CsH) adduct of the methyl group of toluene. This energy difference is also found in the TS for 1,2-addition and in the products, thus leading to a selective activation at the aromatic positions over the aliphatic ones. For the aromatic positions, the TSs for CsH cleavage are of similar energies for para and meta positions, while that at the ortho position is 4.5 kcal mol-1 higher. This is also in agreement with the relative amounts observed experimentally (ortho, 1%; meta, 47%; para, 33%). The reactions of CH4 with CpW(NO)(CH2), Cp*W(NO)(CH2), and the isoelectronic CpW(CO)(CH2)were computed at the B3LYP level by Fan and Hall.184 The main purpose of this work was to test whether an oxidative addition/reductive elimination pathway was possible as an alternative to the 1,2-addition. The d2 electronic configuration in the W-alkylidene reactive species allows, in principle, for a CsH activation through oxidative addition, contrary to the situation with d0 Ti complexes. Even though, the hydrido-alkyl intermediate CpW(NO)(H)(CH3)(CH2),

CsH Bond Activation in Transition Metal Species

Chemical Reviews, 2010, Vol. 110, No. 2 765

Figure 26. Reactivity of Cp*M(NO)(CH2tBu)2 toward CsH bonds.

Cp*W(NO)(H)(CH3)(CH2), and CpW(CO)(H)(CH3)(CH2)could be located on the PES, only in the case of the anionic carbonyl complex was a TS for oxidative addition from a σ(CH4) adduct located. In that case, the oxidative addition/ reductive elimination pathway is slightly preferred over the 1,2-addition. For the nitrosyl complexes, the high energy of the oxidative addition intermediate with respect to the TS for 1,2-addition and the absence of a TS for oxidative CsH cleavage leave the 1,2-addition as the sole mechanism. The authors pointed out that the 1,2-addition is strongly assisted by the metal as illustrated by the short W · · · H contact in the TS (ca. 1.8 Å). The preference for 1,2-addition with nitrosyl is ascribed to the greater π-acidity of this ligand in comparison to CO. In the case of Cp*Mo(NO)(CH2CMe3)2, reaction in benzene at room temperature yielded a mixture of Cp*Mo(NO)(CH2CMe3)(Ph) and Cp*Mo(NO)(Ph)2 in a sequential manner.185 The mixed product Cp*Mo(NO)(CH2CMe3)(Ph) can transform into two different intermediates upon R-migration, Cp*Mo(NO)(CHCMe3) and Cp*Mo(NO)(η2-C6H4), which have been trapped by pyridine.186 B3LYP calculations have been carried out to locate the extrema along the pathway for CpMo(NO)(CH2R)(Ph) reacting with benzene (R ) H, CMe3).186 R-H migration to form the benzyne adduct is significantly more difficult than that to form the alkylidene Mo(CHR). However, the energy difference between the two TSs is significantly influenced by the nature of R (∆∆Eq ) 4.1 kcal mol-1, R ) H; ∆∆Eq ) 10.8 kcal mol-1, R ) CMe3) because the TS for R-migration toward the alkyl is stabilized upon changing R from H to CMe3. In this process, a MsC bond is broken, and this is easier for a secondary carbon (CH2CMe3) than for a primary carbon (CH3), thus showing the importance of representing the experimental system properly in the computational models. From the π-arene complex CpMo(NO)(CHR)(η2-C6H6), the CsH 1,2-addition is computed to be more difficult (∆Eq ) 12.7 kcal mol-1, R ) H; ∆Eq ) 11.5 kcal mol-1, R ) CMe3) than the CsH activation from the benzyne complex CpMo(NO)(η2-C6H4)(η2-C6H6) (∆Eq ) 7.5 kcal mol-1). In all cases, the distance between the migrating proton and the Mo center in the TS is significantly shorter than the sum of the van der Waals radii (Mo · · · H ) 1.856 Å, R ) H; Mo · · · H ) 1.814 Å, R ) CMe3; Mo · · · H ) 1.856 Å, benzyne). This suggests some metal participation in the process.

Figure 27. Reactivity of (PCP′)Ru(CO)(NH2) toward HsH and CsH bonds.

2.3.2. Late Transition Metals New synthetic strategies have now made possible the isolation and study of a variety of monomeric late metal alkoxide, aryloxide, and amide complexes.187 The MsX complexes typically exhibit nucleophilic reactivity, and even deprotonate relatively weak acids. The five-coordinate parent amido complex (PCP)Ru(CO)(NH2) (PCP ) 2,6(CH2P(CMe3)2)2C6H3) reacts with phenylacetylene to yield the Ru(II) acetylide complex (PCP)Ru(CO)(CtCPh) and ammonia.188 The same complex also activates HsH to yield ammonia and the hydride (PCP)Ru(CO)(H). B3LYP calculations have been carried out by Cundari et al. on the reaction of (PCP′)Ru(CO)(NH2) (PCP′ ) 2,6-(CH2PH2)C6H3) with H2 and CH4 (Figure 27).188 Only thermodynamic aspects were evaluated, and the computed values for the reaction with H2 are in agreement with the experimental observations. For the hypothetical CsH activation of methane, the calculations indicate that the reaction is both endothermic and endoergic. This unfavorable thermodynamics of transformation has been ascribed to a large and unfavorable change in RusN bond dissociation energy upon conversion from Ru-amido to Ru-amine (∆BDE ) 40 kcal mol-1). Forming NsH and RusC bonds at the expense of the strong CsH bond only results in a gain of 30 kcal mol-1, not enough to compensate the weakening of the RusN bond. In the reaction with H2, the RusH bond, which is ca. 25 kcal mol-1 stronger than the RusCH3 bond, renders the transformation both exothermic and exoergic. This ML5 d6 complex with a square-based pyramidal geometry could potentially have been a good candidate to promote CsH activation through 1,2-addition. There is no need to create a vacant site because the system is coordinatively unsaturated. Moreover, the d6 configuration in this geometry prevents π-donation from N, so that the lone pair on nitrogen is fully available to abstract a proton. However, the product of the reaction is too unstable to lead to any CsH activation. Evidence for the net addition of arene CsH bonds across a Ru(II)sOH bond was disclosed by Gunnoe, Cundari, and

766 Chemical Reviews, 2010, Vol. 110, No. 2

Figure 28. Selectivity in H/D exchange in toluene catalyzed by TpRu(PMe3)2(OH).

co-workers.189,190 Heating the complex TpRu(PMe3)2(OH) in C6D6 at 80 °C results in H for D exchange at the hydroxide ligand to produce TpRu(PMe3)2(OD). The complex TpRu(PMe3)2(OH) catalyzes H/D exchange between H2O and C6D6. Aliphatic CsH bonds are also activated as illustrated by the results of the reaction with deuterated toluene (Figure 28).191 However, the exchange occurs selectively at the aromatic positions and, in particular, at the para and meta sites. B3LYP calculations on the model (Tab)Ru(PH3)2(OH) (Tab ) HB(NdNH)3-) showed that the reaction proceeds through initial loss of phosphine followed by benzene π-coordination. From this precursor, the CsH activation step is a 1,2-addition across the Ru-O bond to yield a phenyl-water complex (∆Gq ) 17.6 kcal mol-1 and ∆G ) -13.3 kcal mol-1). However, due to the endoergic character of benzene coordination (∆G ) 14 kcal mol-1), the overall transformation (Tab)Ru(PH3)2(OH) + C6H6 f (Tab)Ru(PH3)(H2O)(Ph) + PH3 is also endoergic (∆G ) 18.4 kcal mol-1) and would not lead to any observable Ru(Ph) derivative. However, H for D exchange is still possible as observed experimentally. Interestingly, the phenyl complex (Tab)Ru(PH3)2(Ph) also lies at higher energy (∆G ) 9.1 kcal mol-1), thus avoiding any deactivation pathway in the catalytic H/D exchange process. The H for D exchange is also observed with the anilido complex TpRu(PMe3)2(NHPh), whereas for the methyl complex TpRu(PMe3)2(Me), H for D exchange is preferentially observed at the Tp 4-positions.191 B3LYP calculations on (Tab)Ru(PH3)(Me)(η2-C6H6) yielding (Tab)Ru(PH3)(η2CH4)(Ph) gave ∆Gq ) 21.2 kcal mol-1 for the CsH cleavage step, while it is 17.6 kcal mol-1 with the hydroxide complex (Tab)Ru(PH3)(OH)(η2-C6H6). This lower activation barrier is due to the basic properties of the OH ligand thus stabilizing the transfer of the hydrogen as a proton. In the case of the activation by the methyl group in (Tab)Ru(PH3)(Me)(η2C6H6), the activated CsH bond needs to be closer to the metal (Ru · · · C ) 2.35 Å, Ru · · · H ) 2.02 Å, OH; Ru · · · C ) 2.23 Å, Ru · · · H ) 1.72 Å, Me), and the CsH bond breaking occurs later on the PES (C · · · H ) 1.39 Å, OH; C · · · H ) 1.49 Å, Me).191 For TpRu(PMe3)2X (X ) amido, hydroxo), the 1,2addition of a CsH bond across RusX potentially offers a way to achieve catalytic heterofunctionalization of alkanes or arenes. Gunnoe, Cundari, and co-workers have studied computationally the CsH activation of benzene by [(Tab)M(PH3)2)X]q (X ) OH, NH2; q ) -1 to +2; M ) Tc(I), Re(I), Ru(II), Co(III), Ir(III), Ni(IV), Pt(IV)).192 The CsH activation process is best described as a metal-mediated proton transfer in which the metal center and the X ligand function as an activating electrophile and intramolecular base, respectively. The atoms in molecules (AIM) analysis of the TS for CsH cleavage yielded a picture in agreement with the above characteristics, as bond critical points were only found along the wedges of the kite-shaped four-membered ring. Three criteria have been considered to quantify the efficacy of (Tab)M complexes toward CsH activation and their

Balcells et al.

Figure 29. Reactivity of Ir-methoxo complex toward CsH bond.

potential use as catalysts: (1) generation of the active species (dissociation of the phosphine ligand); (2) hydrogen transfer barrier; (3) thermoneutrality of the hydrogen transfer. A good catalyst is likely to be associated with easy generation of the active species and facile H-transfer, while not generating a too-stable (or too-unstable) product for subsequent functionalization to occur. The complex (Tab)Co(PH3)2(OH)+ is thus predicted to be the most promising for further catalyst investigation with phosphine dissociation associated with ∆G ) 8 kcal mol-1 and CsH cleavage with ∆Gq ) 21.2 kcal mol-1 and ∆G ) 2 kcal mol-1. Concomitantly with the studies carried out by Gunnoe and Cundari, Periana and Goddard developed the CsH activation chemistry with an iridium-methoxy complex (Figure 29).193 The pyridine methoxy complex catalyzes the H for D exchange between mixtures of C6H6 and D2O at 160 °C. All experimental evidence (kinetic studies, KIE evaluation) points to the rate-determining step being benzene coordination followed by fast CsH cleavage. B3LYP calculations on the hydroxo-pyridine complex confirmed the experimental observations.194 Generation of the active species with the hydroxo cis to the vacant site is achieved in two steps: dissociation of pyridine (∆G ) 32.8 kcal mol-1) and isomerization of the chelating ligands (∆Gq ) 10.1 kcal mol-1) amounting in total to an activation barrier of 42.9 kcal mol-1. The activation barrier for CsH cleavage is ∆Gq ) 7.7 kcal mol-1 and the associated TS is 5 kcal mol-1 lower than that for the ligand isomerization. The overall transformation from the hydroxo-pyridine iridium complex to the phenyl-pyridine product is exoergic by -6.8 kcal mol-1, which stands in marked contrast with the Ru complex studied by Gunnoe and Cundari, where a similar transformation is computed to be endoergic by 9.1 kcal mol-1. Oxgaard et al. have further analyzed the 1,2-addition across the IrsOMe bond by computing the localized orbital using the Pipek-Mezey methodology.195 They showed that one lone pair on oxygen forms the OsH bond, while the orbital on O involved in IrsO bonding is converted into the lone pair on HOMe used to coordinate to Ir. The electrophilic metal activates the CsH bond by generating a positively charged hydrogen, which is subsequently abstracted by a basic site on a vicinal ligand (OMe here). Such a transformation was shown by the authors to be different from classical σ-bond metathesis, and they have coined the name internal electrophilic substitution (IES).

2.4. Beyond Classical Descriptions CsH activation processes are generally divided in two extreme mechanisms: oxidative addition and σ-bond metathesis. As illustrated in section 2.1, oxidative addition consists of two steps: coordination of the σ(CsH) bond after generation of a vacant site and actual cleavage of the CsH bond (Figure 1). This transformation is most commonly observed for low-valent electron-rich metals with d8 configuration that allow easy access to an oxidized d6 configuration accommodating two extra ligands.

CsH Bond Activation in Transition Metal Species

Chemical Reviews, 2010, Vol. 110, No. 2 767

Figure 30. Various intermediates involved along a σ-CAM process.

Figure 31. CsH activation as a σ-CAM process between agostichydride and bis-dihydrogen ruthenium complexes.

When no d electrons are available on the metal, an oxidative transformation is not possible and the CsH activation proceeds through a σ-bond metathesis pathway as described in section 2.2 (Figure 1). Essential to such a transformation are the absence of an intermediate and the concerted RsH/MsR′ bond-breaking and R′sH/MsR bond-making processes. For transition metal fragments with d4 to d8 configurations, a σ-bond metathesis transformation is in principle possible, with the additional potential formation of a stable σ-adduct before and after the actual metathesis (Figure 30). Perutz and Sabo-Etienne have coined the term σ-CAM for such a transformation where metathesis is assisted by formation of stable σ-complexes (CAM ) complex assisted metathesis).196 It is usually a multistep process, and one important aspect of the σ-CAM mechanism is that the oxidation state of the metal center remains constant throughout all the steps. This allows design of catalytic cycles at constant oxidation state, as shown by Perutz and Sabo-Etienne for H for D exchange, hydrogenation, and borylation of alkanes.196 In particular, rotation of the σ-bond in the σ-complexes (a situation not possible with classical σ-bond metathesis) opens isomerization pathways leading eventually to functionalization (Figure 30). One early example of a σ-CAM transformation characterized computationally is the CsH activation in a phenylpyridine (ph-py) ruthenium hydride (Figure 31).197 B3PW91 calculations on Ru(PH3)2(H)(H2)(ph-py)+ located the agostic complex featuring coordination of the phenyl ortho-CsH bond, the bis-dihydrogen product, which is less stable by 6.8 kcal mol-1, and the TS for CsH activation (∆Eq ) 25.0 kcal mol-1). Based on the geometrical parameters, the TS was classified as a Ru(IV) species. The complexes involved in the transformation always comprise two coordinated σ-bonds (CsH and HsH bonds in the reactant and two HsH bonds in the product). Reorganization among these σ-complexes through metathesis allows the system to adapt to the stereoelectronic requirements of the ligands (trans influence): hydride trans to σ(CsH) in the reactant and H2 trans to aryl in the product. The ruthenium complex TpRu(PPh3)(CH3CN)H catalyzes the H for D exchange between CH4 and deuterated organic solvents (benzene, THF, Et2O).198 The reaction mechanism was studied computationally by Lin et al. at the B3LYP level with the model TpRu(PH3)(CH3CN)H reacting with C6H6, CH4, and CH3CH2OEt.198 H for D exchanges between CH4 and C6H6 and between CH4 and Et2O were considered as indicated in the catalytic cycle shown in Figure 32. The transformations 7 f 8, 8 f 9, 10 f 11, and 11 f 12 are σ-CAM processes, and rotation of the HD ligand affords H for D exchange between RsD and CH4. The

Figure 32. Catalytic cycle for H for D exchange in the reaction of RsD and CH4 with [Ru](H)(CH3CN) ([Ru] ) TpRu(PH3); R ) C6H5, CH3, CH2CH2OEt).

Figure 33. Selected bond distances (Å) for the TS of CsH activation in the reaction of RsH with TpRu(PH3)(H). Only the atoms in the equatorial plane are shown.

rotation of H2 in TpRu(PH3)(R)(η2-H2) is computed to be very easy (∆Gq ) 1.9 kcal mol-1, R ) Ph; ∆Gq ) 3.8 kcal mol-1, R ) CH2CH2OEt; ∆Gq ) 2.3 kcal mol-1, R ) CH3). From the σ-complexes, the activation barriers for CsH activation are computed to be between 7 and 12.4 kcal mol-1 (∆Gq ) 7.0 kcal mol-1, Ph; ∆Gq ) 11.9 kcal mol-1, CH2CH2OEt; ∆Gq ) 12.4 kcal mol-1, Me), and the Ru(R)(η2H2) product is always less stable than the Ru(η2-RH)(H) reactant (∆G ) 0.8 kcal mol-1, Ph; ∆G ) 5.8 kcal mol-1, CH2CH2OEt; ∆G ) 7.2 kcal mol-1, Me). The most intriguing aspect of this transformation is the geometry of the TS for H-transfer between R and H, featuring a pentagonalbipyramidal geometry with the transferring hydrogen within bonding distance to Ru (Figure 33). The geometry of the transition state is in agreement with a description of the latter as a Ru(IV) complex, hence the name oxidatively added transition state (OATS) proposed by the authors to describe the transformation. The close contact between the transferring hydrogen and the metal makes this reaction very similar to a two-step oxidative addition/reductive elimination where the oxidized species would be an intermediate. Lin et al. have studied computationally the competition between the two pathways for Tp(PH3)M(R) reacting with methane (R ) H and CH3; M ) Fe, Ru, Os).199 When R ) H, the oxidized species Tp(PH3)M(H)2(CH3) is a TS for H migration for M ) Fe or Ru, while it is an intermediate for M ) Os. The situation is almost identical for R ) CH3 with Tp(PH3)Fe(H)(CH3)2 as a TS for H-transfer and Tp(PH3)Os(H)(CH3)2 as an intermediate. The only difference is observed for M ) Ru where a very shallow minimum is found for Tp(PH3)Ru(H)(CH3)2, but the intermediate lies at an energy close to that of the TS for CsH oxidative addition. This shows that the one-step and two-step mechanisms are not mutually exclusive. Any factor stabilizing the higher oxidation state may turn a onestep into a two-step process. This is the case when going down the Fe group. Ancillary ligands may also play a role

768 Chemical Reviews, 2010, Vol. 110, No. 2

(R ) H vs R ) CH3 for M ) Ru). For instance, changing Tp for Cp and computing the H-transfer for CpRu(PH3)R (R ) H or CH3) shows that the CsH activation is a twostep process with a Ru(IV) intermediate. In the context of hydroarylation catalysis (see section 4.2 for further details and references), Oxgaard et al. have studied the CsH activation of benzene by cis-(acac)2Ir(CH2CH2Ph) and TpRu(CO)(CH2CH2Ph).200,201 These complexes result from the ethylene insertion into the MsPh bond. The H-transfer from benzene to MsCH2CH2Ph constitutes the last step of the catalytic cycle, regenerating the active species (see section 4.2). In the case of the iridium complex, the activation barrier for the CsH cleavage from the π-coordinated benzene is computed to be ∆Gq ) 12.9 kcal mol-1, and the resulting σ(CsH) adduct of ethylbenzene is isoenergetic with the benzene complex (∆G ) -0.2 kcal mol-1). Based on the geometry of the TS (Ir · · · C ) 2.27 Å, Ir · · · Ph ) 2.09 Å, Ir · · · H ) 1.58 Å, C · · · H ) 1.69 Å, Ph · · · H ) 1.99 Å), the CsH activation is described as a one-step H-transfer with an Ir(V) transition state. The authors called this process oxidative hydrogen migration (OHM).200 Transition states of similar nature had already been mentioned in the literature by Clot,197 Lin and Eisenstein,199 and Hall.184 The transition state for the hydrogen migration from benzene to CH2CH2Ph in TpRu(CO)(CH2CH2Ph)(η2-C6H6) ressembles that for cis-(acac)2Ir(CH2CH2Ph) described previously (Ru · · · C ) 2.27 Å, Ru · · · Ph ) 2.17 Å, Ru · · · H ) 1.61 Å).201 This TS is described as a Ru(IV) structure with an essentially fully formed RusH bond and partially formed RusC and RusPh bonds as judged by the vibrational frequencies (RusH ) 2110 cm-1, RusC ) 479 cm-1, and RusPh ) 1033 cm-1). Because oxidation to Ru(IV) is generally more difficult than oxidation to Ir(V), the energy barrier is computed to be larger for Ru (∆Hq ) 12.0 kcal mol-1, Ir; ∆Hq ) 18.0 kcal mol-1, Ru). Bercaw showed that σ-bond metathesis of para-substituted benzene is insensitive to the electron-withdrawing or electrondonating property of the para-substituent.136 DeYonker et al. addressed the influence of para-substituent in the activation of C6H5X (X ) Br, Cl, CN, F, H, NH2, NO2, OMe) by TpRu(L)(CH3) (L ) PMe3, CO).202 The computed Ru · · · H distances at the TS for H-migration from the para position of C6H5X to CH3 fall in the range 1.608-1.677 Å (L ) CO) and 1.592-1.620 Å(L ) PMe3), thus classifying the TS as a Ru(IV) hydride. A linear Hammett correlation is calculated with a positive F value of 2.6 for L ) CO and 3.2 for L ) PMe3. Electron-withdrawing para-substituents lower the energy of the transition state. An opposite effect on the energy would have been obtained had the reaction been an electrophilic aromatic substitution. The picture evolving from the present study is that coordination of the substrate to TpRu(L)R promotes CsH bond activation by intramolecular proton transfer. The correlation between the acidity of the CsH bond to be broken and the energy barrier for CsH cleavage was probed by B3LYP calculations on TpRu(PMe3)(CH3) reacting with various substrates (MeNO2, acetone, CH3CN, C6H12, THF).203 The values for the CsH cleavage step (after CsH coordination to Ru) indicate that the more acidic CsH bonds of MeNO2, CH3CN, and acetone are associated with lower ∆Hq values (7.2, 9.8, and 14.8 kcal mol-1, respectively) than the less acidic CsH bonds of cyclohexane and THF (21.5 and 17.2 kcal mol-1).

Balcells et al.

Figure 34. Ligands L considered in the reaction of (L)Pt(CH3)+ with benzene.

Thermal CsH activation of benzene with cationic platinum-methyl complexes was studied in the gas phase by Schwarz et al. (Figure 34).204 The exchange process (L)Pt(CH3)+ + C6H6 f (L)Pt(C6H5)+ + CH4 with the ligands L shown in Figure 34 was effective only for L ) phen, bipyrm, and bipy, because with L ) MeDABDCP chlorine blocks the coordination site for benzene. For the cases where methane was produced, labeling experiments showed that the CsH activation of benzene prior to methane elimination is reversible. DFT calculations were carried out for the reaction of (bipy)Pt(CH3)+ with C6H6, and two different pathways were tested. Oxidative addition of benzene would lead to a ML5 (bipy)Pt(CH3)(C6H5)(H)+ intermediate, which would achieve the exchange upon CsH reductive elimination of H and CH3. Another possibility is to achieve the exchange in one step from the coordinated benzene complex through a σ-CAM process. With the B3LYP functional, the transition states forming the methane adduct along both pathways are computed at a higher energy than the separated product (bipy)Pt(C6H5)+ + CH4. This is not in agreement with the experimental results where reversible CsH activation for benzene was observed. Computation of the two pathways with the mPW1K functional correctly positioned the two TSs below the exit channel. Moreover, the σ-CAM mechanism is favored over the oxidative addition/reductive elimination by 4.1 kcal mol-1. Other functionals were tested and mPW1PW91 and M05-2X yielded correct energies in agreement with the H for D exchange studies. The situation was, however, less clear-cut for the preferred pathway. mPW1PW91 favors σ-CAM, and the two pathways are comparable with M05-2X. A recent experimental and computational study by Bercaw et al. probed the competition between CsH bond activation of benzene and olefin insertion in a (monomethyl)palladium β-diketiminate complex.205 The σ-CAM transition state was computed (B3LYP) to be 9.6 kcal mol-1 lower than the TS for CsH oxidative addition of benzene. The differentiation between the various pathways for CsH activation is often made by inspection of the geometry of the transition state. A short M · · · H contact between the metal and the migrating hydrogen is taken as an indication of M · · · H bonding interaction. Hall et al. have recently proposed to use AIM analysis206 to characterize the bonding paths within the TS for CsH activation.207,208 Seven different molecular graphs were considered for the various possible degrees of connectivity in a four-membered ring geometry (Figure 35).208 Graph 13 is typical of a σ-bond metathesis TS where the transferring hydrogen presents bond critical points with the two ligands between which it is exchanged. The latter are still connected to the metal and the fourcentered TS features a ring critical point. For the TS of the reaction of CH4 with Cp2ScCH3, which is the prototypal σ-bond metathesis reaction, graph 13 is obtained. Graph 19 is typical of the intermediate obtained in the reaction between

CsH Bond Activation in Transition Metal Species

Chemical Reviews, 2010, Vol. 110, No. 2 769

Figure 35. Various molecular graphs obtained with AIM analysis of some CsH activation transition states or intermediates. Figure 37. The rebound mechanism postulated for CsH oxidation by cytochrome P450.

2.5. Rebound Mechanism

Figure 36. The three classes of regime for CsH activation as deduced from ALMO-EDA analysis of the TS.

CpIr(PH3)(CH3)+ and CH4. This oxidative addition intermediate presents the bond critical points between the metal and each of the three ligands. No interaction between the ligands is observed in the AIM analysis. Interestingly, whereas initially the names of the various mechanisms were proposed essentially because a close contact was obtained between M and H, Hall et al. made a connection between specific graphs and mechanisms.207 Thus graph 14 was attributed to metalassisted σ-bond metathesis (MAσBM) as proposed by Hall,184,209 graph 15 was attributed to OATS or σ-CAM, graph 17 was described as a TS for oxidative addition, while finally graph 19 corresponds either to an OHM TS or to an oxidative addition intermediate. Graphs 16 and 18 were not given any specific names. Very recently, Ess et al. have performed B3LYP calculations of the transition states for various CsH activation processes (Figure 36).210 They have used Head-Gordon absolutely localized molecular energy decomposition analysis (ALMO-EDA)211 to partition interaction energies between the metal/ligand and CH4 fragments in the TS geometries. Charge transfer contributions from the metal to CH4 (Ect1) and from CH4 to the metal fragment (Ect2) were evaluated. Depending on the relative magnitude of the two contributions, three different regimes of CsH activation are proposed: electrophilic (Ect2 > Ect1), ambiphilic (Ect1 = Ect2), and nucleophilic (Ect2 < Ect1). See Figure 36 for particular examples of systems for each category.

The oxidation of hydrocarbon compounds by metal-oxo complexes212–214 is one of the most relevant reactions in the field of CsH activation. In nature, several enzymes are capable of oxidizing hydrocarbons with very low reactivities under mild conditions in a catalytic fashion.215 For instance, methane, the most inert alkane, is oxidized to methanol by methane monooxygenase.216,217 In addition, more complex substrates are oxidized in a selective manner, as in the oxidation of camphor by cytochrome P450.218 The mechanism of this reaction, known as the oxygen rebound mechanism (Figure 37), was postulated by Groves.34,219 This mechanism involves three steps: (1) oxidation of the catalyst to generate a high-valent metal-oxo intermediate, (2) H abstraction from the substrate by the oxo group, leading to a metal-hydroxo species and a carbon radical, and (3) OH rebound, in which the hydroxo ligand is transferred to the radical yielding the reaction product and recovering the catalyst. In most cases, the selectivity of the reaction is controlled by the relative stability of the carbon radicals. In saturated substrates, tertiary positions are preferentially oxidized over secondary positions. In unsaturated substrates, vinylic and benzylic positions are preferentially oxidized over saturated positions. Nevertheless, these simple rules can fail in enzymatic systems in which the protein environment surrounding the active site orientates the substrate, promoting the reaction in an unexpected less reactive position. This approach has recently been implemented with success in synthetic manganese catalysts, which promote selective CsH oxidation by molecular recognition.220 Since the rebound mechanism was proposed for cytochrome P450 more than 25 years ago, considerable details on the nature of each reaction step and active species have been unveiled by experimental studies.221–223 This knowledge has been complemented by theoretical studies,224,225 in some cases carried out in collaboration with experimental groups. The simple picture of Figure 37 is now known to be much more complex, with parallel reaction pathways that may involve different active species in different spin states. It has also been argued that the carbon radical may be trapped by the metal before the OH rebound step. In addition, the radical nature of the reaction has been debated. Some studies suggested that the cleavage of the CsH bond may be heterolytic, thus involving hydride transfer and the formation of a carbocation intermediate. The theoretical studies on cytochrome P450 are reviewed by other authors of this special issue, and they are thus not mentioned here.50,51 Nevertheless, the non-heme iron biomimetics of cytochrome P450226,227 are reviewed here.

770 Chemical Reviews, 2010, Vol. 110, No. 2

The unique feature of the rebound mechanism is that it does not involve a direct interaction between the metal and the substrate. The cleavage of the CsH bond is promoted by an oxygen center bound to the metal. Therefore, the metal plays an indirect role by activating the oxygen ligand that oxidizes the substrate. In most cases, the metal induces radical electrophilic character in the oxygen reactive center, which promotes a nucleophilic attack of the CsH bond. On the other hand, if the carbon radical generated by H abstraction binds to the metal, the reaction can be formally regarded as a 1,2-addition. The oxygen rebound mechanism is nowadays accepted not only for cytochrome P450 but for many other enzymes, like methane monooxygenase216,217 and lipoxygenase.228–230 In addition, the rebound mechanism has been also postulated for inorganic reagents like permanganate and the Fenton’s reagent,231 as well as for many biomimetic systems. Due to this diversity, it is now accepted that the nature of the oxidant intermediate is not limited only to metal-oxo species but can also be metal-hydroxo, metal-peroxo, metal-hydroperoxo, bimetallic µ-oxo, and metal-alkoxo species. Theoretical chemists reacted to this diversification by extending their investigations beyond cytochrome P450 to other chemical systems of high interest. The theoretical study of these systems is challenging due to several issues. The proper description of the reactivity of the metal core of the catalyst may require the consideration of several electronic states of different spin. In some cases, these states have strong multireference character, which makes the application of the most commonly used computational methods, like DFT,232 challenging. In addition, some catalysts contain two or more high-spin metal atoms, which are magnetically coupled. In the enzymatic systems, a proper modeling of the reactive center is crucial, and in some cases, QM/MM methods233 must be used to include the influence of the protein environment. Despite these difficulties, computational studies have been successfully used to clarify some key points, like the electronic nature of the active species, the effect of the axial ligand trans to the reactive group, the influence of the pH on the reactivity, the thermodynamic and kinetic parameters of the elementary steps, the influence of the environment, and the kinetic isotope and tunneling effects.

3. Chemically Induced Selectivity 3.1. Thermodynamic Properties: Bond Energy Enthalpies Determination Bond enthalpies are very important quantities to know because they provide key information on the thermodynamic feasibility of a reaction. Reaction enthalpies can be estimated as a sum of bond enthalpy contributions from bonds that are formed and broken. This information is particularly vital for organometallic catalysis where organic molecules are manipulated by transition metal fragments and bonds are cleaved and formed between the metal fragment and organic groups. As mentioned by Landis,234 Shilov’s groundbreaking work in the activation of hydrocarbons was motivated by the assumption that metal-carbon and metal-hydrogen bond enthalpies are reasonably similar.235 It is also important to acknowledge that rational design of hydroamination catalysts resulted from the knowledge of the relative MsN and MsC bond enthalpies.236 In contrast to the large body of information accumulated for organic molecules, experimental studies of transition

Balcells et al.

metal complexes are rare. Furthermore, bonding is often well localized between two atoms in main group chemistry, and thus bond enthalpies are moderately sensitive to substituents; “typical values” can thus be extracted. The situation is quite different for transition metal complexes. Determination of the absolute values of ligand-metal bond enthalpies in polyatomic transition metal complexes is exceptionally difficult and can be achieved only when demanding criteria are met.237–239 In addition, the influence of ancillary ligands on metal-ligand bond enthalpies is unclear but often important, so that the transferability of bond enthalpies between species is questionable. An approach that has been very useful is the determination of relative bond enthalpies in place of absolute values. This gives often the desired information since many reactions consist of interchanges of ligands. Even in this case, assumptions are needed. For instance, if entropic factors and solvent effects are assumed to be constant, bond enthalpies for MsX bonds may be estimated relative to a standard. Bryndza and co-workers were probably the first to notice that a series of relative MsX bond enthalpies correlates with HsX bond enthalpies.240 Their studies included a large variety of X groups, which were bonded to the metal via C, O, or N, and M was either an early transition metal like Sc and Th or a late transition metal like Ru, Ir, and Pt. This very useful discovery was followed by several studies where the same type of correlation was examined focusing on either a given metal fragment or a limited choice of X groups. Two studies focused on hydrocarbyl groups. Through elaborate procedures employing kinetic and equilibrium studies, it was possible to obtain a bond energy correlation where the experimental values of the MsC bond enthalpies relative to a reference value were plotted against the HsC experimental bond dissociation enthalpies. These studies were carried out for the case of the RhsC bond of Rh(H)(R)(Tp*)(CNCH2CMe3) (R ) hydrocarbyl, and Tp* ) HB(3,5-dimethylpyrazolyl)3) by Jones and co-workers241,242 and for the TisC bond of Ti(R)(silox)2(NHSitBu3) (silox ) OSitBu3, R ) various alkyl or aryl groups) by Wolczanski and co-workers.173,243 Another experimental study concerned the correlations between NisN and NisO bond enthalpies and the corresponding HsN and HsO bonds.244,245 Experimental correlations were also determined for cyclopentadienyl complexes of Zr and Hf with a wide variety of ligands including hydride, hydrocarbyl, alkoxide, amide, and halide.246 The scarce experimental information available was complemented by several computational studies. Here again, the contrast between organic and organometallic chemistry is impressive. Computational study of bond dissociation enthalpies in main group molecules is a well-documented field. The size and nature of the organic molecules that have been considered have permitted the use of highly elaborate methods, and accurate bond enthalpies values have been obtained.247–252 The size and nature of transition metal complexes made the accurate calculation of MsX bond enthalpy more difficult. Ab initio calculations were used notably by Goddard, Siegbahn, and Bauschlicher to calculate MsH and MsR bond energies in naked systems, that is, with no other ligand than H or R. DFT methods were required for larger complexes containing ligands, and some of the shortcomings of these methods have been discussed.250,253,254 However these methods were used with success for the study of relative bond energies in a large variety of complexes.

CsH Bond Activation in Transition Metal Species

The relative MsH and MsCH3 bond energies for a variety of naked and CO-coordinated metal complexes, M(CO)5R (M ) Mn, Tc, Re) and M(CO)4R (M ) Co, Rh, Ir), were studied, and periodical trends were discussed.255 Further studies included other carbon-based ligands256 and cyclopentadienyl complexes.257 It was, for instance, established that the MsCH3 bond is considerably weaker than the MsH bond for late to middle transition metal as a result of repulsive interaction between occupied orbitals on the methyl and metal fragments. Isomeric alkyl complexes of early and late transition metals were studied by Harvey.258 The alkyl R groups cover a large range from primary to tertiary groups. It was shown that primary alkyl complexes are usually more stable than secondary or tertiary ones due to electronic effects. The metal-carbon bond is often substantially ionic in nature with a large negative charge on the alkyl group and the stability of carbanions decreases from primary to tertiary groups. For related reasons, electron-withdrawing substituents at the β-position increase the stability of the complexes. The steric effects, which are often invoked, are shown to play only a minor role. A systematic study of metal-ligand bond enthalpies for saturated transition metal complexes that encompasses the entire d-block of the periodic table with a wide assortment of ligands has been carried out.234 To minimize the steric effects, all ancillary ligands were set as hydrides, so the calculations of the MsL bond enthalpy by DFT and also wave-based methods such as CCSD(T) have been carried out on HnMsR (R ) H, non-hydrocarbyl, substituted hydrocarbyl, etc). This study shows that large variations in metal-ligand bond enthalpies across a row undermine the utility of intrinsic bond enthalpies. However MsH and MsCH3 bond enthalpies increase down a period in accord with experimental values. The relation between relative HnMsR bond enthalpies and the corresponding HsR bond enthalpy was considered for individual metals. The best fit is obtained when R is restricted to a family of groups such as hydrocarbyl. The slope of the correlation line between calculated HsC and MsC bond enthalpies, RM-C/H-C, computed for individual metals, varies between 1.2 and 1.9 over the periodic table with larger values obtained for early metals. The large values for early metals coincide with metal-carbon polarization. It is, however, not possible to describe the large variation in the metal-carbon enthalpies by Pauling’s formula. A study of the RM-C/H-C bond energy correlations was carried out using computational models that represent closely experimental systems for which the correlation was experimentally studied.259 The correlations RM-C/H-C of the bond enthalpies, established for Rh and Ti complexes by Jones and Wolczanski, respectively, were reproduced with high accuracy. With DFT(B3PW91) calculations, Rh(H)(R)(Tp)(CNMe) to model Rh(H)(R)(Tp*)(CNCH2CMe3), and including all R groups that have been used in the experimental studies, a correlation was obtained for which the slope of the linear fit is 1.23, which agrees very well with the experimental value of 1.22. A similar study was carried out for the titanium complexes family studied experimentally by Wolczanski. In this case, the full experimental system, Ti(R)(silox)2(NHSitBu3), was included in the model. The calculated RM-C/H-C ratio of 1.08 agrees very well with the experimental value of 1.12. It was concluded that while absolute bond enthalpies are reproduced within 5% for HsC

Chemical Reviews, 2010, Vol. 110, No. 2 771

bonds, the very limited data for absolute MsC bond enthalpies do not allow an accurate comparison between experiment and theory. In contrast, computational evaluation of HsC and MsC relative bond enthalpies allows the prediction of trends and changes in relative bond enthalpy with substituents. The same sets of molecules were calculated with the BP86 method and analyzed with the energy decomposition method (EDA) to determine the steric contribution to the bond energy.260 The results are similar to those presented by Clot.259 The slope is slightly different (1.10 and 0.96 for Rh and Ti, respectively) and slightly away from the experimental values. The steric effects are found to be important within the set Me, Et, Pe, iPr and tBu. The experimental observation that MsC bond energies increase more rapidly with changes in the substituents than HsC bond energies has considerable consequences on the selectivity of the activation of the HsC bond. Wick and Jones summed it up as follows: “The ground state energy differences between alkyl and aryl hydride complexes of rhodium dominate the reactivity. The difference in MsC bond strengths is larger than the corresponding difference in CsH bond strengths giving rise to a strong thermodynamic preference for cleaving strong CsH bonds.”242 The experimental measurements of Jones and Wolczanski show that the MsC bond strength increases about 10-20% faster than the HsC bond one. Computations were thus used to determine whether higher increases in the MsC bond energy could be obtained.261,262 The fluoroarenes HsC6FnH5-n (n ) 0-5) were selected because fluorine has a strong influence on the nearby HsC bond energies. For instance, computations show that in C6H5F, the weakest CsH bond is at the meta position, the para CsH bond is 0.9 kcal mol-1 stronger, and the ortho CsH is 2.5 kcal mol-1 stronger. Similarly, the CsH bond dissociation energy of C6F5H, 114.1 kcal mol-1, is 4.1 kcal mol-1 higher than that of benzene. The selected systems for the DFT(B3PW91) studies are shown in Figure 38. They are models for complexes that are known to be able to activate CsH bonds or able to form aryl complexes. They include metals having formal electron counts varying from d0 to d8, and they also represent a wide variety of coordination spheres including M(η5C 5 H 5 ) 2 (H)(Ar F ), M(η 5 -C 5 H 5 ) 2 (H) 2 (Ar F ), M(η 5 5 C5H5)(H)(ArF)L2, M(η -C5H5)(H)(ArF)L and M(H)(ArF)L2. Figure 39 gives the results for RhCp(PH3)(H)(ArF). The variations in the MsC bond dissociation energies relative to that of Msphenyl (∆D(RhsC)rel) are plotted versus the corresponding CsH bond dissociation energies for the fluorinated benzenes (∆D(HsC)rel). A linear correlation appears with a slope RM-C/H-C of 2.43 ( 0.11 (r2 ) 0.981). Remarkably, the role of the fluorine substituent strongly depends on its position on the benzene ring, since the 20 values form themselves into three subsets corresponding to zero ortho F (the lowest values for ∆D(RhsC)rel), to one ortho F, and two ortho F substituents (the highest values for ∆D(RhsC)rel). A very similar pattern was obtained for all metal fragments shown in Figure 38 with RM-C/H-C varying between 1.9 and 3 (Figure 40). As already noted by Landis, the values of RM-C/H-C do not appear to follow any clear pattern with the nature of the metal center. The highest values of RM-C/H-C have been found for ZrCp2(H)(ArF) and for Ni(dhpe)(H)(ArF). However, some pattern appears with the ligands: increasing the

772 Chemical Reviews, 2010, Vol. 110, No. 2

Balcells et al.

Figure 40. RM-C/H-C for Zr ) ZrCp2(H)(ArF), Ta ) TaCp2(H)2(ArF), Ta′ ) TaCp2Cl(H)(ArF), W ) WCp2(H)(ArF), Re ) ReCp(CO)2(H)(ArF), Re′ ) ReCp(CO)(PH3)(H)(ArF), Re′′ ) ReCp(PH3)2(H)(ArF), Rh ) RhCp(CO)(H)(ArF), Rh′ ) RhCp(PH3)(H)(ArF), Ir ) IrCp(CO)(H)(ArF), Ir′ ) IrCp(PH3)(H)(ArF), Pt ) Pt(dhpe)(H)(ArF), and Ni ) Ni(dhpe)(H)(ArF).

Figure 38. Metal-ArF (n ) 0, 5) for which the correlation factor RM-C/H-C was calculated.262

Figure 39. Correlation between the RhsC bond dissociation energies of Rh(η5-C5H5)(H)(ArF)(PH3) and the HsC bond dissociation energies of fluorobenzenes. The zero corresponds to the values for HsPh and RhsPh bond energies.

electron density at the metal by increasing the donating ability of one ligand (Cl vs H or CO vs PH3) results in an increase of RM-C/H-C. Interpreting the correlations has met difficulties similar to those mentioned by Landis. The best results were obtained by correlating ∆D(MsC)rel with the total negative charge of the aryl ring, qaryl, although qaryl is more influenced by meta and para F than ∆D(MsC)rel. Thus, although the Pauling model for bonding provides a qualitative framework for interpreting the results, other factors play a role. The large influence of ortho fluorine was attributed to the delocalization of the σ lone pair of the phenyl anion into the adjacent empty σ*(CsF) orbitals. The large value of RM-C/H-C computed for fluoroarene was experimentally confirmed in the case of Tp*Rh(CNneopentyl)(ArF)H with

a very good agreement between the experimental and calculated values for RM-C/H-C of 2.14 and 2.0, respectively.263 The influence of F-substitution on the kinetics of reaction was studied in the case of the reaction of CpRe(CO)2 with fluoarenes.109 In this reaction, the ratelimiting step is the isomerization from the cis isomer, which is the product for the CsH activation, to the thermodynamic trans product. The energy barriers of this isomerization were found to correlate with the Re-C bond dissociation energies. Increasing the number of fluorines, notably at the ortho position, lowers the energy barriers. In this reaction, thermodynamic and kinetic aspects favor cleavage of the strongest CsH bond. The relation between kinetics and thermodynamics was also studied for the intramolecular CsX bond activation in Ru(NHC)(PH3)2(CO), where NHC ) 1-(C6H4-2X)imidazol-2-ylidene and X ) H, CH3, F, OH, NH2, OCH3, or CF3.264 The computed exothermicity of C(aryl)sX activation follows the trend X ) NH2 < CH3 < H < OH = OCH3 < CF3 < F, while the energy barriers vary as X ) H < F < OH = OCH3 < CF3 < NH2 < CH3. Both series reflect the promotion of C(aryl)sX activation by the formation of stronger bond in the product.

3.2. Chelation-Assisted CsH Activation 3.2.1. ortho-CsH Activation Even though CsH bonds do not form very stable σ-adducts with transition metals, coordination of the bond that will break is a compulsory event prior to actual cleavage. Thermodynamics of the MsC bond formed favors aromatic CsH activation over an aliphatic CsH one, and in the latter the primary carbon reacts preferentially over the secondary and tertiary positions. However, within a substituted aromatic molecule, little selectivity is generally observed between the various CsH bonds and usually steric arguments are used to rationalize the results obtained. In the case of fluorinated aromatics, activation of the CsH bonds maximizing the number of ortho F atoms is strongly favored thermodynamically (see section 3.1). One major breakthrough in the field of aromatic CsH functionalization has been achieved by Murai with olefin insertion into a CsH bond ortho to a carbonyl group in aromatic ketones (Figure 41).16,18 The reaction is generally

CsH Bond Activation in Transition Metal Species

Figure 41. Ru-catalyzed olefin insertion into the ortho CsH bond of an aromatic ketone. [Ru] ) RuH2(CO)(PPh3)3.

performed in refluxing toluene with RuH2(CO)(PPh3)3 as a catalyst precursor. The active species is thought to be a Ru(0) complex, Ru(CO)(PPh3)3 or Ru(CO)(PPh3)2, and the CsC coupling has been shown experimentally to be the ratedetermining step. B3LYP calculations have been performed by Morokuma on benzaldehyde and Ru(CO)(PH3)2 as models for the experimental system.265 O-coordination to Ru of the aldehyde is preferred over π-coordination of the arene by 11.9 kcal mol-1. This coordination of the orientating group yields an agostic complex, which opens easy access to the selective activation of the ortho CsH bond of benzaldehyde. A transition state associated with this coordination has been located with ∆Eq ) 3.3 kcal mol-1. From the agostic intermediate, the CsH cleavage is very easy (∆Eq ) 2.8 kcal mol-1) and the product is 11.1 kcal mol-1 more stable than the agostic precursor. It is interesting to note that in the process of ortho-CsH activation with coordination of the carbonyl group, the most difficult step of the CsH activation is not CsH cleavage but the formation of the agostic. The crucial role played by the coordination of the orientating group is illustrated by the calculations of the ortho-CsH cleavage from the π-arene complex, without any participation of the carbonyl group. The energy barrier is computed to be ∆Eq ) 20.1 kcal mol-1. CsH activation at the meta and para positions, for which no coordination of the remote carbonyl group is possible, are calculated to be easier (∆Eq ) 18.0 kcal mol-1, meta; ∆Eq ) 17.8 kcal mol-1, para). For the CsH activation at the ortho position, there is an energy difference of 29.4 kcal mol-1 between the two transition states for CsH cleavage (with and without coordination), thus showing the essential role played by the orientating group in achieving selectivity. Chaudret and Sabo-Etienne have shown that ethylene insertion into the ortho-CsH bond of acetophenone catalyzed by Ru(H)2(H2)2(PCy3)2 is effected at room temperature (20 °C).266 Products of ortho-CsH activation of acetophenone and phenyl-pyridine were isolated, but in the case of phenyl-pyridine, no ethylene insertion was observed (Figure 42). Protonation of the phenyl-pyridine orthometalated

Chemical Reviews, 2010, Vol. 110, No. 2 773

species yielded a σ(CsH) adduct, considered to be an intermediate in the Murai reaction.197 B3PW91 calculations by Clot showed that the CsH activation from this σ(CsH) adduct to yield a bis-H2 intermediate is best described as a σ-CAM process through a TS structure with a short Ru · · · H contact (see section 2.4).197 The H2 ligand trans to the RusC bond is labile and can be substituted by another L ligand, in particular the solvent used in the synthesis (THF). The influence of the nature of L on the energetics of the σ-CAM process has been tested computationally.267 When no ligand is present trans to the forming Ru · · · C bond, there is nearly no energy barrier for the CsH cleavage (∆Eq ) 0.1 kcal mol-1). If the ligand trans to σ(CsH) is Me2O, as a model for the experimental solvent THF, the energy barrier is still very low (∆Eq ) 1.5 kcal mol-1). Interestingly, when the cis-hydride, receiving the migrating H atom in the σ-CAM process, is positioned trans to the σ(CsH) bond and thus a vacant site is cis to the bond to cleave, the energy barrier increases to a high value of ∆Eq ) 29.4 kcal mol-1. This can be attributed to the destabilization introduced by putting two strong σ-donor ligands in trans geometry (hydride and forming RusC), but it also shows that the CsH cleavage is eased by the cis-hydride. ortho-CsH activation of aromatic ketones was also observed by Esteruelas.268,269 Treatment of OsH6(PiPr3)2 with acetophenone and 2-fluoroacetophenone in toluene under reflux affords the osmium trihydride complexes resulting from ortho-CsH activation (Figure 43).268 Similar ortho-CsH activation was obtained when the starting complex was a dehydrogenated phosphine stannyl complex (Figure 43).269 In the latter, deuterium labeling experiments suggest that the ortho-CsH activation takes place from OsH(SnPh2Cl)(PiPr3)2. The reaction pathway for the formation of this intermediate has been evaluated by DFT calculations.269 The studies by Chaudret and Esteruelas show that, contrary to the case studied by Murai, Ru(II) and Os(II) complexes can be active species in ortho-CsH activation. The results with 2-fluoroacetophenone indicate that the stabilization induced by coordination of the orientating group to selectively promote ortho-CsH activation is larger than the stabilization induced by activating a CsH bond ortho to a fluorine atom. The ortho-fluorine effect (see section 3.1) has been explained in terms of increase of the ionic component of the MsC bond through inductive effect of the ortho-F atom. Milstein has extended this influence to other σ-attracting groups such as Cl, Br, and OMe. Selective ortho-CsH

Figure 42. ortho-CsH activation reactivity observed between Ru(H)2(H2)2(PCy3)2 and acetophenone or phenyl-pyridine.

774 Chemical Reviews, 2010, Vol. 110, No. 2

Balcells et al.

Figure 43. ortho-CsH activation promoted by osmium polyhydride complexes.

Figure 44. Selective ortho-CsH activation of haloarenes by an Ir(I) system.

Figure 45. Kinetic CsH and thermodynamic CsCl oxidative additions with Ir(I)-PNP pincer complex.

activation of haloarenes and anisole was observed with an Ir(I) complex (Figure 44).270,271 The product is always a mixture of activation at the ortho, para, and meta positions, with ortho-CsH activation yielding the major kinetic product. Upon prolonged heating at 50 °C, the ortho-CsH activation species is recovered as the sole thermodynamic product (para activation is still observed with fluorobenzene). X-ray structures for the product of chlorobenzene and anisole ortho-CsH activation present short Ir · · · X distances (Ir · · · Cl ) 2.816 Å, Ir · · · OMe ) 2.76 Å). The observed selectivity is a result of coordination of the heteroatom to the metal center, which kinetically directs the metal to the ortho-CsH bond and stabilizes the resulting complex thermodynamically. DFT(mPW91) calculations were carried out on the model system (PNP)Ir(coe)+ (PNP ) 2,6-bis(dimethylphosphinomethyl)pyridine, coe ) cyclooctene) and chlorobenzene or anisole.271 The CsH activation is effected from the 14electron intermediate, (PNP)Ir+, whose formation upon coe dissociation is the rate-determining step for the entire process. Calculations have identified various η2-(CdC) and η2-(CsH)

adducts of C6H5X (X ) Cl, OMe) to the iridium center and the transition states interconverting these adducts. The preferred pathway consists of η1-coordination of the heteroatom, followed by isomerization to the ortho η2-(CsH) complex, thus leading to selective activation. After CsH cleavage, coordination of X in the remaining vacant site is shown to bring 6-10 kcal mol-1 extra stabilization, thus reinforcing the thermodynamic stability of the ortho-CsH activation product. Differences in reactivity were observed by Ozerov with another PNP-iridium complex (Figure 45).272 The reaction between the iridium dihydride and norbornene in C6H5Cl produced a mixture of four CsH oxidative addition products (two regioisomers for activation at the meta positions, one for activation at ortho and para) and a small fraction of the CsCl oxidative addition product. Thermolysis at 70 °C did not change the amount of the CsCl oxidative addition product but converted all the CsH activation products into the ortho-CsH activation isomer as proven by X-ray crystallography (Ir · · · Cl ) 2.96 Å). Finally, thermolysis

CsH Bond Activation in Transition Metal Species

Chemical Reviews, 2010, Vol. 110, No. 2 775

Figure 46. Competition between R- and β-(CsH) activation in the reaction of 2-butanone with Ir.

above 100 °C converted the CsH activation product into the CsCl activation isomer, thus indicating that the latter is the thermodynamic product for this system. Similar behavior was observed with bromobenzene. The competition between CsH and CsCl activation in chlorobenzene by (PNP′)M (PNP′ ) bis(Z-2-(dimethylphosphino)vinylamino); M ) Ir, Rh) as a model for the experimental system was explored with B3LYP calculations by Wu and Hall.273,274 In the present system, no Clcoordination to Ir is involved to selectively activate the orthoCsH bond. Five different products of CsH activation of chlorobenzene were located on the PES depending on the relative position of the hydride and the chlorine atom (two ortho, two meta, and one para isomers). The CsH cleavage is shown to proceed from a π-arene complex, and despite the lower stability of the π-adduct at the C(1)dC(2) position (with respect to π-adduct at other positions on the ring), the ortho-CsH activation product with Ir · · · Cl interaction is ca. 7 kcal mol-1 more stable than the four other CsH activation products. In the case of the (PNP)Ir+ system studied by Milstein and Martin, the product of CsCl activation was computed to be 9.5 kcal mol-1 less stable than the product of ortho-CsH activation.271 In the (PNP′)Ir neutral system, the product of CsCl activation is 22.5 kcal mol-1 more stable than the product of ortho-CsH activation. The TS for direct CsCl oxidative addition from an η1-Cl complex is 7.4 kcal mol-1 higher in energy than the TS for CsH oxidative addition of the ortho-CsH bond. The thermodynamic product (PNP′)IrCl(C6H5) is obtained from the ortho-CsH activation kinetic product through CsCl oxidative addition forming a benzyne intermediate. The works of Ozerov and Hall indicate that coordination of the orientating group before the CsH cleavage step is not mandatory to obtain selective activation at the ortho position. However, the stability of the ortho-activated product is significantly enhanced by coordination of the orientating group after CsH cleavage. This has been shown experimentally by Goldman with the reaction of nitrobenzene with (PCP)IrH2 (PCP ) κ3-C6H3-2,6(CH2PtBu2)2) in the presence of norbornene.275 The experimental results are best explained if CsH activation is effected prior to O-coordination of the nitro group. Moreover, as the kinetic selectivity is quantitative for the meta and para CsH bonds, the ortho-CsH bond activation is kinetically hindered rather than assisted by proximity of the coordinating group. The same authors studied the reaction of nitromethane with (PCP)Ir yielding the bidentate O,O-ligated nitromethanate complex (PCP)Ir(H)(κ2-O,O-NO2CH2).276 PBE calculations by Krogh-Jespersen showed that CsH activation is assisted by chelation of one oxygen atom and yields (PCP)Ir(H)(κ2O,C-NO2CH2) as a kinetic intermediate, which rearranges to the observed product.276 Selective sp3 CsH activation of ketones at the β-position was observed by Milstein with (PNP)Ir+ (Figure 46).277 Activation at the R-position was observed only when water was present. DFT calculations on the actual experimental system were carried out at the COSMO-PBE0 level.

The rate-determining step, deciding for the outcome of the reaction, was shown to be the isomerization from the O-coordinated ketone to the η2-(CsH) adduct prior to CsH cleavage. Because of the steric bulk imparted by the phosphine substituents, the O-coordinated complex with the methyl group of 2-butanone toward Ir is 5 kcal mol-1 more stable than with the ethyl group toward Ir. As the former leads to R-activation and the latter to β-activation, the observed product (β-activation) is favored because of destabilization of the ground state. After CsH activation, coordination of the carbonyl further stabilizes the system toward activation at the β-position (five-membered ring) but destabilizes the product of R-activation (four-membered ring). However, upon coordination of water, the product of R-activation is almost isoenergetic with the product of β-activation through creation of a six-membered ring exhibiting a hydrogen bond between the water and the ketone group of the substituted alkyl ligand. The thermal reaction of Tp*Ir(C6H5)2(N2) with 2-ethylphenol produces the hydride alkylidene as the main product (Figure 47).278 The hydride alkene isomer that derives from β- as opposed to R-hydride elimination in the last CsH activation step amounts to only 5% of the total reaction product. Lewis bases like NCMe induce a stereospecific 1,2 hydrogen shift (Figure 47). ONIOM(BHandH:UFF) calculations of the various intermediates shown in Figure 47 are in agreement with the experimental observations. The hydride alkylidene complex is 1.8 kcal mol-1 more stable than the hydride-alkene. Reaction of Tp*Ir(C6H5)2(N2) with anisole generates an heteroatom-stabilized hydride carbene (Figure 47).279,280 DFT(BHanH) calculations on the actual experimental system were carried out to determine the sequence in the three consecutive CsH activations, that is, C(sp2)sH before or after C(sp3)sH. Initial interaction between iridium and anisole is by way of O-coordination. Forming the σ(CsH) adduct with either the ortho-C(sp2)sH or with one methyl C(sp3)sH bond does not introduce a particular energetic bias (∆E ) 10.1 and 9.6 kcal mol-1 from the O-coordinated isomer for C(sp3) and C(sp2), respectively). In both cases, the CsH cleavage is effected through a σ-CAM transformation generating a benzene complex, and the energy difference between the sp2 and sp3 CsH activation TS is only 0.5 kcal mol-1. Substitution of the coordinated benzene by the CsH bond yet to be activated is endothermic by less than 1 kcal mol-1. A second σ-CAM process leads to the CsH activation of the other type of bond, and in both cases, the TS for the second CsH activation is at a lower energy than for the first one by ca. 5 kcal mol-1. Dissociation of benzene generates a vacant site cis to the Ir-CH2 group, thus affording the observed hydrido alkylidene product after R-CsH activation. From the initial O-coordinated anisole, the reaction is exothermic by 2.8 kcal mol-1, and no particular sequence (sp2 and sp3 vs sp3 and sp2) is preferred.

776 Chemical Reviews, 2010, Vol. 110, No. 2

Balcells et al.

Figure 47. Hydride alkylidene and hydride alkene Ir isomeric complexes formed from CsH activation of 2-ethylphenol.

Figure 48. Zirconium-catalyzed coupling of propene and R-picoline.

3.2.2. CsH Activation in N-Heterocycles Predictable methods for heterocycle elaboration are critical to the synthesis of pharmaceuticals. The ring heteroatom of an N-heterocycle can promote activation of an adjacent CsH bond by coordination to the metal center.20 The first example of a direct catalytic coupling between a heterocycle and an olefin was reported by Jordan and Taylor in 1989 (Figure 48).281 The catalytic cycle proposed by Jordan (Figure 48) was studied computationally (B3LYP) by Lin et al.282 The last step of the catalytic cycle is the regeneration of the η2-picolyl complex from the Cp2Zr(H)(R-picoline) complex. This transformation (24 f 25 in Figure 48) is a typical σ-bond metathesis with ∆Gq ) 19.3 kcal mol-1 and ∆G ) 5.4 kcal mol-1. The σ-bond metathesis process involving a CsH bond of the methyl group is computed to be more difficult (∆Gq ) 21.6 kcal mol-1) and less favored thermodynamically (∆G ) 7.4 kcal mol-1). However, the difference is not very large

and CsH activation of the methyl group is possible, as illustrated by deuterium incorporation when D2 is used. The observation of only one product is due to the relative energies of the transition states for propene insertion. The reaction at the orthometalated carbon in 25 is favored by ∆∆Gq ) 16.2 kcal mol-1 over the insertion of propene at the ZrsCH2 bond in 22. Hydrogenolysis of the ZrsC bond in 22 (Figure 48) is also a σ-bond metathesis, and the reaction is computed to be the rate-determining step with ∆Gq ) 25.3 kcal mol-1 and ∆G ) -5.4 kcal mol-1. Substitution of the functionalized R-picoline by a new R-picoline and CsH activation regenerate the active species. This two-step process could have been achieved in one-step through a direct σ-bond metathesis between R-picoline and the ZrsC bond of 22. Transition state structures for the one-step process (with and without explicit N-coordination of the incoming R-picoline) were located, but their energies were very high (∆Gq ) 48.8

CsH Bond Activation in Transition Metal Species

Figure 49. Catalytic CsH activation of a substituted benzimidazole.

and 41.7 kcal mol-1, respectively). The easier pathway mediated by H2 is due to the nondirectional nature of the 1s orbital on H involved in the metathesis reaction, thus lowering the energy of the corresponding TS (see section 2.2). The thorium alkyl complex Cp*2Th(Me)2 and R-picoline react to give preferentially C(sp3)sH bond activation in the presence of a more reactive C(sp2)sH bond, while the analogous uranium complex Cp*2 U(Me)2 reacts with only the ortho-CsH bond of R-picoline.283,284 B3LYP calculations were carried out on the competition between C(sp2)sH and C(sp3)sH activation in R-picoline mediated by Cp2An(Me)2(An ) Th, U).285 The reaction begins with the formation of a weakly bound N-adduct of R-picoline. This is followed by a σ-bond metathesis of either the ortho-CsH bond or one methyl CsH bond with one of the two methyl ligands. Activation at the sp2 position is thermodynamically favored for the two actinide metals. The relative activation energies between C(sp2)sH and C(sp3)sH bond activation differ slightly between Th (∆Eq(sp2) > ∆Eq(sp3)) and U (∆Eq(sp2) < ∆Eq(sp3)). This is in agreement with the experimental observations where sp2 activation is the thermodynamic product for both metals, while sp3 activation is the kinetic product in the case of Th. Metal-mediated CsH bond activation reactions have become important methods for the formation of carbon-carbon bonds. Ellman and Bergman have proposed Rh(I)-catalyzed cyclization of aromatic and heterocyclic compounds through CsH activation followed by intramolecular coupling to an alkene (Figure 49).286,287 The currently accepted mechanism for this type of cyclization is to start with aromatic CsH activation to form an intermediate Rh-hydride complex. Insertion of the alkene into the RhsH bond is followed by CsC reductive elimination to yield the cyclic product. In the present case, Elmann and Bergman showed that a substantially different mechanism is operative. N-heterocyclic carbene (NHC) complex 26 was shown to be a critical intermediate. This NHC complex presents the same catalytic activity as [RhCl(coe)2]2/ PCy3, the catalyst normally used, thus implying that the NHC complex is the resting state of the catalyst. DFT(B3LYP) calculations were conducted on a model system (see Figure 49) to estimate the energetics along the cyclization pathway from the NHC intermediate.286

Figure 50. Mechanism of carbene formation via CsH activation.

Chemical Reviews, 2010, Vol. 110, No. 2 777

Even though the catalysis is performed at 135 °C to overcome the high barrier associated with alkene insertion, the formation of the NHC resting state via CsH activation is achieved under much milder conditions (THF, 75 °C). A detailed investigation of the mechanism of the NHC-complex formation was carried out on 3-methyl-3,4-dihydroquinazoline and the catalyst system [RhCl(coe)2]2/PCy3 (Figure 50).287 The N-coordinated intermediate 29 was shown to be a precursor of the NHC complex and the calculated structure for this complex (PH3 instead of PCy3) identified a short Rh · · · H contact (2.7 Å) defined as “pre-agostic”.288 The C2deuterated substrate reacted with the usual Rh(I)/PCy3 mixture to produce a N-deuterated NHC complex with 88% isotope incorporation at N1, thus indicating that the C2-H bond of the reactant is the primary source of N1-H in the NHC complex. A double-labeling crossover experiment proved that the C2-H to N1-H transformation is intramolecular. The intramolecular pathway for the CsH activation was characterized computationally (B3LYP) with Rh(PMe3)2Cl as a model of the catalyst and 3-methyl-3,4-dihydropyrimidine as a model of the reactant. The high-energy process is the isomerization of the system from the N-coordinated complex (observed experimentally) to a σ(CsH) complex of the C2-H bond. The N-coordination thus selects the CsH bond to be cleaved. Migration of Rh affords a different, yet isoenergetic, σ(C2-H) complex from which CsH cleavage is easily achieved and yields a hydride intermediate. This intermediate presents the required syn geometry to form the NsH bond of the NHC complex. The proposed mechanism is in agreement with the experimental observations and the calculated activation parameters (∆Gq ) 28.3 kcal mol-1, ∆Hq ) 24.0 kcal mol-1, and ∆Sq ) -14.5 cal K-1 mol-1) are in excellent agreement with the experimental values (∆Gq ) 29.1 kcal mol-1, ∆Hq ) 26.0 kcal mol-1, and ∆Sq ) -10.3 cal K-1 mol-1). In a series of papers, Esteruelas and co-workers have studied C(sp3)sH activation in substituted pyridines promoted by osmium complexes (Figure 51).289–293 DFT(B3LYP) calculations by Lledo´s on OsH6(PMe3)2 estimated the energy cost of H2 elimination to form OsH4(PMe3)2 at 29.3 kcal mol-1.289 N-coordination of 8-methylquinoline and 2-(dimethylamino)pyridine stabilized the osmium tetrahydride by 9.4 and 13.2 kcal mol-1, respectively. For the CsH activation to proceed, a vacant site is necessary. This is achieved by H2 dissociation to generate an osmium dihydride featuring an agostic methyl. From this agostic intermediate, CsH activation is energetically easy (∆Eq ) 3.1 kcal mol-1, 8-methylquinoline; ∆Eq ) 0.4 kcal mol-1, 2-(dimethylamino)pyridine) and the process is a classical oxidative addition. The osmium tetrahydride product is more stable than the N adducts (∆E ) -2.8 kcal mol-1, 31; ∆E ) -4 kcal mol-1, 32). The DFT calculations identified the products of the protonation of 31 and 32 as bis-dihydrogen complexes.289 These bis-H2 complexes lose H2 and evolve through Hmigration to the hydrido-dihydrogen osmium complexes with an agostic methyl interaction (see Figure 51). The process

778 Chemical Reviews, 2010, Vol. 110, No. 2

Balcells et al.

Figure 51. Reactivity observed with osmium polyhydrides and various N-heterocycles (P ) PiPr3).

is a σ-CAM transformation, and it is easier for 8-methylquinoline than for 2-(dimethylamino)pyridine (∆Eq ) 13.3 kcal mol-1 and ∆Eq ) 16.9 kcal mol-1, respectively). In the case of 8-methylquinoline, the product of the H-migration is 3.8 kcal mol-1 more stable than the bis-dihydrogen complex and corresponds to the observed product 33. For 2-(dimethylamino)pyridine, the product of H-migration is less stable than the bis-dihydrogen complex by 2.5 kcal mol-1. One dihydrogen ligand can easily dissociate, and the remaining H2 undergoes an oxidative addition process to yield the osmium dihydride 35 with an R-agostic interaction of the Os-CH2 group. From this intermediate, oxidative addition yields easily the product 34 observed experimentally (∆Eq ) 9.7 kcal mol-1 and ∆E ) -3.9 kcal mol-1). Experimentally, upon protonation, 32 is transformed in a carbene complex 34 as the result of double CsH activation. The mechanism of NH-tautomerization of R-picoline and quinoline on osmium has been studied computationally (Figure 51).290–292 N-coordination to OsH2Cl2(PMe3)2, followed by hydride transfer from osmium to nitrogen to generate an η2-(CdN) pyridinium complex is associated with a high activation barrier (∆Gq ) 48.9 kcal mol-1, R-picoline; ∆Gq ) 39.3 kcal mol-1, quinoline). An intermolecular hydrogen transfer between OsH2Cl2(PMe3)2 and R-picoline or quinoline without N-coordination presents a lower barrier (∆Gq ) 23.6 kcal mol-1, R-picoline; ∆Gq ) 23.9 kcal mol-1, quinoline). CdN η2-coordination of the pyridinium formed opens a pathway for CsH activation at the R-position. In the case of R-picoline, the rate-determining step is the CsH activation, while it is the H-transfer from Os to N in the case of quinoline. The η2-(CdN) pyridinium complex isomerizes to an η2-(CRsH) complex where the CsH bond is aligned with the P-Os-P vector. It is thus perpendicular to the equatorial plane where the two chloride ligands and the hydride are situated. A σ-CAM process is thus not possible even if the product obtained (37 and 38) would have

suggested it. CsH cleavage is effected through an oxidative addition process generating a seven-coordinated osmium(IV) dihydride complex with a reduced P-Os-P angle (ca. 130°). This complex isomerizes to the Os(II) dihydrogen product (37 or 38) observed experimentally, and the trans geometry for the phosphine ligands is recovered. Overall the NH tautomerization leading to the osmium dihydrogen carbene complex is computed to be significantly exoergic (∆G ) -13.6 kcal mol-1, R-picoline; ∆G ) -15.9 kcal mol-1, quinoline). A recent study has revealed that the aromatic positions of pyridine, 3-methylpyridine, and 4-methylpyridine can be sequentially deuterated in the presence of OsH4(PiPr3)2 by catalytic H for D exchange with benzene-d6.293 The deuteration rates of the pyridine CsH bonds depend on their position in the heterocycles. B3PW91 calculations were carried out on OsH4(PMe3)2 and pyridine or 4-methylpyridine to explain the selectivities observed experimentally. As expected, the CsH activation is composed of two steps: coordination of the CsH bond and actual cleavage. For pyridine, the stability of the η2-(CsH) complexes varies as position 2,6 < position 4 < position 3,5. The relative order for the CsH cleavage step is position 3,5 > position 2,6 > position 4. A combination of both factors yields the para position as the preferred site of deuteration as observed experimentally. As a methyl group impedes coordination of the adjacent CsH bonds, the methyl substituent has a marked negative effect on the deuteration of its adjacent positions. Double geminal CsH activation and reversible R-elimination in 2-(dimethylamino)pyridine Ir(III) complexes has been observed by Crabtree et al. (Figure 52).294 DFT(B3PW91) calculations have been carried out on IrH2(PH3)2+ and 2-(dimethylamino)pyridine to elucidate the factors responsible for the smooth transformation observed in Figure 52. The substitution of two acetone ligands by N-coordinated pyridine and agostic amino-methyl group is only slightly

CsH Bond Activation in Transition Metal Species

Chemical Reviews, 2010, Vol. 110, No. 2 779

Figure 54. Unexpected role of CO in CsH oxidative addition by a cationic Rh(I) complex. Figure 52. Double geminal CsH activation in 2-(dimethylamino)pyridine Ir(III) complex.

endoergic (∆G ) 3.1 kcal mol-1). From the agostic complex, CsH cleavage is easy (∆Gq ) 10.1 kcal mol-1) and gives an iridium dihydrogen hydride complex (∆G ) -1 kcal mol-1). Contrary to the chemistry observed by Esteruelas with neutral osmium, the present cationic iridium promotes CsH activation through a σ-CAM transformation. The TS structure is an Ir(V) trihydride, but two hydrides couple in the product to create an H2 ligand trans to the newly formed Ir-C bond. This reorganization of the nature of the ligands favors a smooth transformation. Subsequent dissociation of H2 generates a vacant site that is needed for the second CsH activation. However, the vacant site is trans to the bond to activate and isomerization to a cis-phosphine geometry affords easy R-elimination (∆Gq ) 2.1 kcal mol-1 and ∆G ) -9.7 kcal mol-1). The cis-phosphine carbene complex then converts to the trans-phosphine product observed experimentally.

3.2.3. CsH Activation in Ancillary Ligands Coordination of an orientating group promotes selective CsH activation in the vicinity of the group as illustrated by the Murai reaction35 and the propene insertion in R-picoline proposed by Jordan.281 In these reactions, the substrate is preconditioned by appropriate chemical substitution to react at a given position. This approach can also be used in ligand design to build new architectures that may lead to more efficient catalytic systems. In this respect, PCP pincer complexes have shown great potential because they are usually very stable at the high temperature required for some catalytic reactions.295 N-heterocyclic carbenes (NHC) stabilized by bulky groups have also been shown to be very good ligands to build efficient catalysts.296 Coordination of the stable free NHC is a common synthetic procedure, but, in some cases, CsH activation of the imidazolium has been shown to be an alternative. Moreover, the bulky groups usually present on the nitrogen atom may be engaged in intramolecular CsH activation processes with the metal. Reaction of the ligand 1,3-bis(di-tert-butyl(phosphino)methyl)benzene (39) with the [Rh(CO)]+ fragment in THF

resulted in clean formation of the bis-chelated complex (40), which contains an η2-(CsH) bond (Figure 53).297 The coordination of the CsH bond increases its acidity as illustrated by the easy formation of the PCP pincer complex 41 upon exposure to NEt3 or collidine. B3LYP calculations on a model of 40 (Me instead of tBu on phosphorus) revealed that the ring bond lengths in the agostic complex are affected only weakly, relative to those of the free ligand. Protonation of the free ligand to yield a Wheland type structure resulted in significantly larger changes in bond distances within the aromatic ring, with clear bond alternation as expected. Thus, a strong agostic interaction between a metal center and an aromatic compound followed by deprotonation can be considered as an alternative to electrophilic aromatic substitution. Milstein and Martin evidenced an unexpected role of carbon monoxide in CsH oxidative addition by a similar cationic Rh(I) complex. (Figure 54).298 In the present system, the CsH oxidative addition is promoted by addition of the π-acceptor ligand CO, which is contrary to expectation for an oxidative addition favored by electron-rich metal centers. Two pathways are proposed for the transformation differing in the order for the two necessary steps, CO coordination and CsH activation, to convert the agostic complex to the hydrido aryl isomer (Figure 54). DFT(PBE0) calculations were carried out to elucidate which of the two mechanisms is preferred. CsH oxidative addition in 42 to generate the hydrido aryl intermediate 44 is associated with an activation barrier ∆Gq ) 11.1 kcal mol-1 and is endoergic (∆G ) 7.8 kcal mol-1). Coordination of CO leads to 43, which is more stable than 42 (∆G ) -1.6 kcal mol-1). Coordination of CO to the agostic complex 42 leads to the agostic dicarbonyl complex 45, which is 1.2 kcal mol-1 more stable than 42. Quite surprisingly, the CsH activation in 45 has a lower barrier than that in 42 (∆Gq ) 9 kcal mol-1), and therefore the TS for CsH activation in 45 lies at 3.3 kcal mol-1 below that for CsH activation in 42. Thus, while the first CO ligand inhibits CsH activation by locking the system with an agostic interaction, the coordination of the second CO ligand facilitates CsH activation. The influence of the strength of the agostic interaction on the competition between aromatic and aliphatic CsH activa-

Figure 53. Formation of η2-(CsH) agostic rhodium arene complexes and their deprotonation.

780 Chemical Reviews, 2010, Vol. 110, No. 2

Balcells et al.

Figure 55. Aromatic vs aliphatic CsH bond activation by Rh(I) as a function of agostic interactions.

Figure 56. Intramolecular CsH activation in a strained SCS Ir pincer complex.

tion has been studied by Milstein and Martin in the reaction of PC ligands with a cationic Rh(I) complex (Figure 55).299 The aryl-PC type ligand 46 yields the CsH activated complex 47. The latter undergoes reversible CsH activation and H for D exchange into the hydride and aryl ortho-CsH with D2O and MeOD. This complex also promotes catalytic H for D exchange into the vinylic CsH bond of olefin. The benzyl-PC ligand 48 yields the CsH activated complex 49, which is very different from 47.300 In addition, 49 is not active in H/D exchange with ROD and in catalytic H for D exchange with olefin. DFT(mPW1k) calculations of the mechanism for CsH activation and H exchange were carried out on the model systems 47t and 49t shown in Figure 55. The main difference between the aliphatic system and its aromatic counterpart is in the activation barriers for the reductive elimination process from the Rh(III)-hydride to the Rh(I)-(η2-(CsH)) agostic complex (∆Gq ) 12.7 kcal mol-1, aliphatic system; ∆Gq ) 4.5 kcal mol-1, aromatic system). A comparison of the geometries of the aliphatic and aromatic agostic complexes, together with Wiberg bond indices, indicates that the agostic interaction is stronger in the aromatic case. Moreover, in the aromatic case, the agostic interaction essentially develops with the carbon and the protonic character of the hydrogen slightly increases, whereas in the aliphatic case, the agostic interaction is essentially with the hydrogen atom and the hydridic character of the latter increases. Thus, the aromatic PC system demonstrates both nucleophilic (oxidative addition) and electrophilic (H for D exchange) type of CsH activation, whereas the aliphatic PC system shows only nucleophilic CsH activation by oxidative addition. Reaction of the bis(diphenyl(phosphino)sulfide)methanide anion with [Ir(coe)2Cl]2 affords a hydrido complex (Figure 56).301 A DFT(B3PW91) study of the mechanism of formation of 51 was carried out on a model system where all the phenyl groups but one are replaced by hydrogen atoms and the coe ligand is modeled by ethylene. The rate-determining

step is the decoordination of the sulfide to allow coordination of one ortho-CsH bond of the phenyl group (∆Hq ) 22.3 kcal mol-1). From this agostic complex, CsH activation is easy and exothermic (∆Hq ) 12.6 kcal mol-1 and ∆H ) -4.8 kcal mol-1) and coordination of the sulfide group yields the product. DFT calculations suggest that only a very weak energy barrier is needed to promote the formation of the Ir(III) hydrido complex 51 through intramolecular oxidative addition of a CsH bond of one pendant aromatic ring. A similar reactivity with a Ru(II) complex has been studied experimentally and computationally.302 The simplest synthetic procedure to form an NHC complex uses free NHC as a reagent. The substituents on the nitrogen atoms are usually large enough to stabilize the carbene. One of the often used ligands is 1,3-dimesitylimidazol-2-ylidene (IMes). However, the mesityl group may be engaged in intramolecular CsC and CsH bond activation as shown by Whittlesey (Figure 57).303,304 Ru(IMes)(PPh3)2(CO)(H)2 (52) reacts with alkenes at room temperature to give the product of intramolecular CsH activation 54. The CsH activation is effected from the Ru(0) intermediate 53 after hydrogen abstraction by the alkene. The reaction is reversible and addition of H2 to 54 regenerates 52. The same complex 52 can also be induced to undergo an alternative intramolecular CsC activation. Heating 52 at 80 °C with excess IMes results in substitution of PPh3 by IMes to generate the dihydride 55. Further heating at 110 °C yields 56, the product of C(sp2)sC(sp3) activation, with release of methane and substitution of IMes by PPh3. The formation of 56 is irreversible. ONIOM(BP86:HF) calculations estimated the energy barriers from the various Ru(0) complexes of PPh3 and IMes either tri- or tetracoordinated. The energy barriers for CsH activation from these complexes are small enough to observe CsH activation at room temperature. However, the energy barriers are significantly lower for the tricoordinated complexes than for the tetracoordinated systems. In contrast, the

CsH Bond Activation in Transition Metal Species

Chemical Reviews, 2010, Vol. 110, No. 2 781

Figure 57. CsC and CsH bond activation reactions in N-heterocyclic carbene complexes of ruthenium.

energy barriers for CsC cleavage are significantly larger and heating is required to overcome the hurdle. The introduction of a second IMes ligand does not promote CsC cleavage. It serves to stabilize the unsaturated Ru(CO)(IMes)2 intermediate and thus makes a reactive intermediate accessible. In this system, CsH activation is always preferred kinetically, but this reaction is reversible and upon prolonged heating, and with excess of IMes, the product of CsC activation accumulates. Recently, a B3LYP mechanistic study has been carried out to explore the structural and energetic features leading to the decomposition pathways of a Grubbs second-generation catalyst.305 The theoretical results strongly suggest that the deactivation pathway initiates with a CsH activation process. Because NHC ligands have often been presented as alternatives to phosphines and are considered to render the metal center more electron-rich, oxidative addition processes are expected to be easier upon substitution of phosphine by NHC. To test this hypothesis, Macgregor has studied computationally (BP86) the CsH activation of methane by Ru(CO)(PH3)n(IR)3-n (n ) 1-3; for R ) H, IR ) (1,3imidazol-2-ylidene); for R ) Me, IR ) (1,3-dimethylimidazol-2-ylidene)).306 The energy barriers are all computed to be ca. 18 kcal mol-1, and the reaction is approximately thermoneutral. Interestingly, the energy barrier increases upon substitution of PH3 by IH or IMe, even though computations confirm that introducing NHC ligands renders the metal more electron-rich. A fragment analysis reveals that increasing the electron density at Ru destabilizes the occupied dπ levels, thus increasing the Lewis basicity of the metal. However, a more important effect is a reduction in Lewis acidity, and this factor lies behind the similar reaction energetics computed for analogous PH3 or IH-containing species. Base-assisted generation of NHC from imidazolium salts is another common strategy for synthesizing NHC complexes when the nitrogen substituents do not afford enough stabilization to the free carbene.307 A combined experimental and computational study of CsH oxidative addition of bisimidazolium salts to iridium complexes has been carried out by Peris and Lledo´s (Figure 58).308 The first metalation affords Ir(I) complexes, and the calculations could not differentiate between the two possible pathways, deprotonation of the imidazolium by the base and coordination or CsH oxidative addition to Ir and deprotonation of the cationic Ir-hydride by the base. The second IrsC bond is formed through CsH oxidative addition as illustrated by the product observed with the short methylene linker (Figure 58). Calculations confirm that for the second metalation, CsH oxidative addition is the preferred pathway over the direct deprotonation of the imidazolium followed by coor-

Figure 58. Formation of chelate NHC complexes through CsH activation.

dination. Because of the geometric constraints imposed by the coe ligand and the already formed MsC bond, three different geometries are possible for the interaction of the second CsH bond to cleave with the metal center. Two out of the three geometries lead to an isomer with cis H and Cl in the product of CsH activation, whereas in the remaining one H and Cl are trans. From these cis isomers, HsCl reductive elimination yields the square planar complexes observed with the longer linkers. In the case of the short C1 linker, the CsH cleavages forming the trans and cis isomer (with respect to the positions of H and Cl) have similar energy barriers, and the more stable trans product is obtained under thermodynamic control. For the longer C3 case, the CsH activation leading to the trans isomer is effected through a TS lying 3 kcal mol-1 higher in energy than the one yielding the cis isomer. The latter is thus formed kinetically and readily eliminates HCl to form the observed square planar bis(NHC)-Ir(I) species. A unified mechanism can apply to the formation of bis(NHC)-Ir(III)-H and bis(NHC)-Ir(I) complexes, based on the oxidative addition of the imidazolium CsH bonds to the metal. The weak base can play the role of an HCl reductive elimination inductor, rather than a deprotonating agent. In a study of the formation of NHC complexes through CsH activation of imidazolium salts, Crabtree observed a striking influence of the counteranion A- that switches the ligand binding from C2 to C5 (Figure 59).309,310 With R ) Me and A- ) Br-, the expected product of metalation at the C2 position (58) is obtained as the major product (91: 9). With A- ) BF4- as the counteranion and R ) Me, a 45:55 mixture of both isomers (C5 isomer, 59, major product) is obtained. Increasing the bulk at nitrogen (R ) iPr) slightly lowers the amount of the C2 product (84:16) in the case of Br and triggers the system toward exclusive formation of the C5 product (59) in the case of BF4-. ONIOM(B3PW91:UFF) calculations have shown that bonding at C2 is thermodynamically preferred for the naked

782 Chemical Reviews, 2010, Vol. 110, No. 2

Balcells et al.

Figure 59. Counterion effects on the geometry of metalation of imidazolium salts to iridium. P stands for PPh3.

cation (∆E ) 10.1 kcal mol-1).310 Calculations of the ionpair structure311 identified the location of the anion-cation contact at the backbone of the pyridine-NHC ligand. This brings extra stabilization by creating H-bonding interactions between the anion and the ligand. Due to the greater acidity of the C2-H bond, the H-bonds are stronger for isomer 59, and the energy difference between 58 and 59 almost cancels (∆E ) 1.6 kcal mol-1, BF4-; ∆E ) 3.8 kcal mol-1, Br-). The computational study of the reaction pathways leading to the C2- and C5-carbene isomers was carried out on a model system consisting of IrH3(PMe3)2 and the N,N′dimethylimidazolium cation in the presence of Br- or BF4-.310 The anion-induced selectivity was shown to originate from different anion participation in the TS for CsH activation. With Br-, a strong H-bond develops with the acidic proton at C2, and this interaction is maintained in the TS during the cleavage of C2-H. The anion follows the hydrogen atom that is transferred as a proton. The transformation is a σ-CAM process as the initial η2-(CsH) bond in interaction with Br is converted in an η2-(HsH) bond still in interaction with Br-. The larger BF4- anion, involved in a network of H-bonds, is not able to stabilize such a σ-CAM transformation, and oxidative addition at the C5 position is computed to be preferred. The reason for the change of mechanism is the very strong donor character of the C5carbene, which favors attainment of Ir(V) versus the highly protonic character of the C2-H of an imidazolium salt that favors the heterolytic CsH path. This relatively rare case of an anion-dependent selectivity change is therefore explained by a relatively rare coincidence of several criteria. In a very recent study, Whittlesey and Macgregor have shown that a base selectively deprotonates the alkyl CsH bond geminal to an agostic interaction.312 With N,N′dimethylimidazol-2-ylidene as a base, calculations (BP86) indicated that the abstraction of the nonagostic proton has an energy barrier of ∆Eq ) 12.7 kcal mol-1. In contrast, reaction profiles for the approach of the external base toward the agostic hydrogen did not lead to deprotonation. The base remains a spectator, and CsH oxidative addition is associated with an energy barrier of ∆Eq ) 19.6 kcal mol-1, leading to an unstable Ru(IV) intermediate (∆E ) 18.9 kcal mol-1). This Ru(IV) intermediate is deprotonated by the base (∆Eq ) 3.6 kcal mol-1) to yield the final product. This computational study indicates that the base-induced cleavage of a nonagostic CsH bond is possible despite the perception that the latter should be less acidic than the agostic CsH bond. DFT(BP86) calculations have been carried out on the cycloplatination reaction of cis-[Pt(Me)2(dmso)(P(o-tol)3] leading to the C,P-cyclometalated compound with liberation of methane.313 The reaction was shown to proceed in four steps: (i) dissociation of dmso, (ii) intramolecular CsH activation of a methyl group on the P(o-tol)3 ligand, (iii) reductive elimination of methane to give the σ-complex, and (iv) fast reassociation of the dmso ligand.

Figure 60. Catalytically selective CsH oxidation of ibuprofen by molecular recognition.

The competition between carbene and olefin products of CsH activation within the PCP ligand tBu2PNHC3H4NHPtBu2 on ruthenium via competing R- and β-H elimination was studied experimentally and computationally by Gusev.314 Double intramolecular CsH activation of a pendant group of a PNP pincer ligand (tBu2PCH2SiMe2)2N) to yield an alkylidene ruthenium complex was studied experimentally and computationally by Caulton.315,316

3.3. CsH Activation under Molecular Recognition High selectivities are often achieved in enzymatic reactions thanks to molecular recognition. In this strategy, a functional group of the substrate binds to the recognition site of the enzyme by means of weak noncovalent interactions, like π-stacking or H-bonds. This prevents the reaction on the functional group and orientates a specific fragment of the substrate toward the reactive center, thus promoting the reaction in a selective manner. Very recently, this approach was successfully implemented in an artificial catalytic system by Brudvig and Crabtree36,220,317 The active catalytic species, [(terpy′)(O)Mn(µ-O)2Mn(SO4)(terpy′)]+, is a mixed-valence dinuclear manganese complex with two bridging oxo ligands (Figure 60). Each metal center is bound to a functionalized terpyridine ligand, and one apical position is occupied by an oxo ligand whereas the other is occupied by a sulfate ligand. The reactivity of this complex was theoretically studied by Eisenstein et al.318 considering the oxidation of toluene to benzyl alcohol as model reaction. The calculations showed that the substrate is oxidized by the apical oxo group of the catalyst in its ground state, which is an antiferromagnetic doublet.319 The reaction follows the classical rebound mechanism, as in the mononuclear manganese systems studied by the same authors (Vide infra).320–322 The CsH oxidation of ibuprofen involves both benzylic positions, and a mixture of two products is thus obtained. Nevertheless, when the reaction is catalyzed by [(terpy′)(O)Mn(µ-O)2Mn(SO4)(terpy′)]+ only the isobutyl group is oxidized. The terpy′ ligand is a terpyridine functionalized with a phenyl ring, used as linker, and a Kemp’s triacid, used as recognition site. The origin of selectivity was explained by postulating the molecular recognition model represented in Figure 60.220 The carboxylic group of ibuprofen is recognized by the catalyst, which prevents the oxidation of the -CH(CH3)COOH fragment. The other substituent of the phenyl ring, -CH2CH(CH3)2, is brought into the reactive site, thus undergoing selective CsH

CsH Bond Activation in Transition Metal Species

Chemical Reviews, 2010, Vol. 110, No. 2 783

Figure 61. Energy diagram (kcal mol-1) associated with the CsH bond activation of RsH by Pd(κ2-O2CH)2 for R ) C6H5, CH3.

oxidation. This model was theoretically confirmed.318 The calculations showed that the substrate is recognized by a double H-bond, which is maintained through the whole reaction pathway. The substrate does not fit perfectly into the pocket open between the recognition and reactive sites, but the system is flexible enough to promote the reaction and preserve the double H-bond. The reaction without molecular recognition, which would be nonselective, is not observed due to the energy penalty of ∼15 kcal mol-1 associated with the loss of the double H-bond. This study also revealed that the catalyst may bind to two molecules of substrate simultaneously. One molecule reacts with the oxo group, whereas the other blocks the approach of unrecognized molecules of substrate to the reactive site, which will diminish the selectivity of the reaction. This model also explained the inhibition of the catalyst by tert-butyl benzoic acid found experimentally.317

3.4. Base-Assisted CsH Activation The Fujiwara reaction represents an early example of CsH functionalization of aromatic compounds through CsH bond activation.15 In this reaction, the catalyst is Pd(OAc)2, and the coordinated acetate is considered to play an active role in the CsH bond cleavage process. A very recent review by Macgregor and Davies addresses specifically the description of the mechanism of CsH activation at late transition metal systems.38 An early computational study of CsH bond activation of benzene and methane by M(κ2-O2CH)2 (M ) Pd, Pt) has been carried out by Sakaki.323 In this work, a comparison between benzene and methane CsH activation by M(PH3)2 has also been conducted. For benzene and methane activation by M(κ2-O2CH)2, the reaction proceeds in two steps (Figure 61). The η2-coordination of the CsH bond requires partial decoordination of one formate ligand. In the intermediate species, the coordinated CsH bond presents a short contact between the hydrogen and the oxygen atom of the formate that is not bonded to M. In the case of M ) Pd, the distances are 1.958 and 2.141 Å for benzene and methane, respectively. As the acidity of the CsH bond increases upon coordination to the electrophilic metal center, these relatively short distances are indicative of extra stabilization. Accordingly, the energy of the intermediate

relative to the separated reactants is computed (MP4SDQ) to be -9.8 kcal mol-1 for M ) Pd and benzene and 8.8 kcal mol-1 for M ) Pd and methane. From this intermediate, the CsH cleavage is effected through proton transfer from C to O, and the energy barrier is low (∆Eq ) 16.1 kcal mol-1, M ) Pd, benzene; ∆Eq ) 11.4 kcal mol-1, M ) Pd, methane). At the TS for H-transfer, the Pd · · · H contact is long showing that there is no Pd(IV) character in this step. It is interesting to compare these results with those obtained for CsH activation by Pd(PH3)2. In this case, the energy barriers for CsH oxidative addition of benzene and methane are 26.5 kcal mol-1 and 34.7 kcal mol-1, respectively. The CsH activation mediated by Pd(II)-formate is thus significantly easier than the CsH oxidative addition at Pd(0)-phosphine. Another particular aspect of the reaction with Pd(O2CH)2 is that the oxidation state of the metal is constant throughout the transformation, whereas with Pd(PR3)2 the activation product is a Pd(II) complex. The reaction is computed to be exothermic for Pd(κ2-O2CH)2 (-16.5 kcal mol-1, benzene; -8.3 kcal mol-1, methane), and endothermic for Pd(PH3)2 (22.1 kcal mol-1, benzene; 31.5 kcal mol-1, methane). The determining factor for the favorable thermodynamics of the CsH activation by Pd(O2CH)2 is the creation of a strong OH bond in the formate ligand. Analysis of the electron distribution indicates that the CsH bond activation by M(κ2-O2CH)2 is heterolytic, while in the case of M(PH3)2, the activation is homolytic. Acetate-assisted CsH activation is also a potential alternative in cyclometalation reactions as illustrated in the works by Davies and Macgregor.324–327 The various pathways (oxidative addition, 1,2-addition, and proton transfer to acetate) for cyclometalation of dimethylbenzylamine (DMBAH) with Pd(OAc)2 were characterized by BP86 calculations (Figure 62).324 In the acetate-assisted CsH activation, the rate-determining step is not CsH cleavage but formation of the agostic intermediate. After coordination of the NMe2 group of DMBA-H to Pd, substitution of one coordinated acetate oxygen atom with one ortho-CsH bond within DMBA-H is associated with an energy barrier of ∆Eq ) 13.0 kcal mol-1. The agostic intermediate presents a short H · · · O contact of 2.04 Å, and the proton transfer to O is almost barrierless (∆Eq ) 0.1 kcal mol-1). In comparison, the TS

784 Chemical Reviews, 2010, Vol. 110, No. 2

Figure 62. Possible transition states for CsH activation of DMBA-H by Pd(OAc)2.

for the 1,2-addition across the Pd-O bond and the TS for oxidative addition have significantly larger energy barriers (34.3 and 25.7 kcal mol-1, respectively). The acetate-assisted CsH activation is thus clearly preferred. After CsH cleavage, the protonated acetate rotates and transfers the hydrogen to the other acetate, creating a hydrogen bond stabilizing the product. It also allows adaption of the electronic influence of the ligands to the new bonding situation. The strong σ-donor aryl ligand is trans to the weak ligand that constitutes the acetic acid bonded by way of the oxygen of the CdO bond. A similar mechanism was reported by Lledo´s for the selective cyclopalladation of R3PsNCH2aryl iminophosphoranes with Pd(OAc)2.328,329 Davies and Macgregor also reported DFT calculations showing that intramolecular hydrogen bonding to acetate can provide a low-energy pathway to CsH activation with iridium (Figure 63).325,326 For the acetate-assisted CsH activation, no agostic intermediate was located, and the TS structure presents a geometry similar to that observed for the Pd-agostic intermediate (CsH ) 1.11 Å, H · · · O ) 2.15 Å).324 The energy barrier of 16 kcal mol-1 is similar to that obtained with Pd (13 kcal mol-1). The cyclometalation is slightly exothermic (∆E ) -2.4 kcal mol-1). Whereas the CpIr fragment is usually associated with CsH activation via oxidative addition, in the present case the oxidative addition TS has a high energy (∆Eq ) 30.7 kcal mol-1). Even though the 1,2-addition, via a four-membered ring in the TS, has a lower energy barrier (∆Eq ) 14.0 kcal mol-1), it starts from

Balcells et al.

a κ1-acetate isomer that is 8.8 kcal mol-1 less stable than the κ2-acetate ground state. Consequently, the TS for the 1,2addition is 6.8 kcal mol-1 above that assisted by acetate, via a six-membered ring (Figure 63). Ligands such as κ2-OAc can provide both strong basic character and, via arm dissociation, a geometrically convenient route for intramolecular H-transfer. DFT(BP86) calculations have been employed to study the role of chelating base, RCO2- (R ) Me, Ph, CCl3, CF3, OH, Ph) in the cyclometalation reaction of DMBA-H with CpIr.326 In contrast to the results of the previous study,325 these new calculations identified a two-step pathway for the acetateassisted CsH activation similar to that observed with Pd(OAc)2.324 The formation of the agostic intermediate with dissociation of one acetate arm has an energy barrier of ∆Eq ) 13.4 kcal mol-1. In contrast to the case of Pd, the interaction of the agostic CsH bond is with the oxygen atom coordinated to Ir and not the one that has dissociated. This would tend to indicate that a 1,2-addition pathway is preferred but the corresponding TS is 11 kcal mol-1 above the agostic intermediate. The TS for H-transfer to the noncoordinated oxygen via a six-membered ring is computed to be only 0.9 kcal mol-1 less stable than the agostic intermediate. The two-step process involving initial κ2-κ1 displacement of the base, followed by facile CsH cleavage, was found for PhCOO- and CCl3COO-, while with HCOOand CF3COO-, no agostic intermediate was located. The calculations show that CsH activation is controlled by the energy of the κ2-κ1 isomerization of the base, so higher energy barriers are computed with more strongly coordinating bases RCOO- (∆Eq ) 13.4 kcal mol-1, R ) Me; ∆Eq ) 14.6 kcal mol-1, R ) Ph; ∆Eq ) 9.5 kcal mol-1, R ) CCl3). The proton transfer from the agostic intermediate has a low energy barrier that is rather insensitive to the actual basicity of RCOO- (∆Eq ) 0.9 kcal mol-1, pKa ) 4.8, R ) Me; ∆Eq ) 0.9 kcal mol-1, pKa ) 4.3, R ) Ph; ∆Eq ) 1.3 kcal mol-1, pKa ) 0.65, R ) CCl3). The basicity of the free base is shown to be significantly attenuated by binding to the metal

Figure 63. Computed reaction profiles (kcal mol-1) for CsH activation in CpIr(DMBA-H)(OAc)+. The product of CsH oxidative addition (top right) is computed to lie 0.4 kcal mol-1 below the associated TS prior to the inclusion of ZPE corrections.

CsH Bond Activation in Transition Metal Species

Figure 64. NsH versus CsH activation of a pyrrole imine at iridium.

Figure 65. Pd-catalyzed intramolecular arylation.

center. The counterdirecting effects of coordination strength and basicity mean that the overall energy associated with CsH activation in these species is remarkably insensitive to the nature of the intramolecular base employed. The (IrCl2Cp*)2/NaOAc system can activate the NsH bond of a pyrole imine at the expense of a CsH bond (Figure 64).327 However, with the N-methylated ligand, CsH activation occurs. The competition between NsH and CsH activation by CpIr(κ2-O2CMe)+ on the model substrate HNdCHNC4H4 was studied by DFT(BP86) calculations. Two isomers of the precursor complex CpIr(HNdCHNC4H4)(κ2-OAc)+ were located with different orientations of the pyrole ring, thus creating NsH · · · O or CsH · · · O interactions. The former situation is more stable than the latter by 2.6 kcal mol-1. In both cases, the TS for XsH cleavage is associated with κ2-κ1 isomerization of acetate, and the H-transfer is easier from X ) N than from X ) C (∆Eq ) 14.2 kcal mol-1, X ) N; ∆Eq ) 19.6 kcal mol-1, X ) C). Thermodynamics is also in favor of NsH activation (∆E ) -0.3 kcal mol-1, X ) N; ∆E ) 1.2 kcal mol-1, X ) C). It is perhaps surprising that the barrier to CsH activation is only 5.4 kcal mol-1 larger than that for NsH activation. This reflects the synergetic nature of acetate-assisted XsH activation, where both XsH · · · O H-bonding and XsH · · · M agostic interactions reinforce each other to facilitate the bond activation process. The effect of the acetic acid as a solvent to assist the cyclometalation of amino-imine in palladium complexes has been studied computationally by Aullo´n and co-workers.330 The mechanism of acetate-assisted CsH bond activation by (acac-O,O)2Ir(OAc) was studied computationally by Goddard.331 Interaction energies for benzene and methane CsH bond activation were evaluated using the ALMO-EDA method. The TS for acetate-assisted deprotonation via a six-membered ring has a significantly lower activation barrier than the TS for 1,2-addition across the IrsO bond via a four-membered TS (∆Gq ) 25.4 kcal mol-1, six-membered ring, ∆Gq ) 44.7 kcal mol-1, four-membered ring, benzene; ∆Gq ) 30.2 kcal mol-1, six-membered ring, ∆Gq ) 49.5 kcal mol-1, fourmembered ring, methane). Ambiphilic bonding character in these transition states is the result of an electrophilic iridium metal interacting strongly with the incoming methyl or aryl group along with a nearly equivalent amount of covalent bonding resulting from deprotonation by acetate.

Chemical Reviews, 2010, Vol. 110, No. 2 785

In the studies by Davies and Macgregor, the CsH bond to activate is brought into close proximity to the acetate group by coordination of the NMe2 group of DMBA-H. This favorable situation can be achieved in intramolecular arylation reactions catalyzed by palladium. Generally, the catalyst system is composed of a source of Pd(0), a bulky phosphine ligand, and a base (Figure 65). CsX (X ) halogen) oxidative addition on Pd(0) anchors the substrate on the metal and generates a situation adapted for further CsH activation. Echavarren and Maseras have studied the proton-abstraction mechanism in the palladium catalyzed intramolecular arylation with a combined experimental and theoretical approach (Figure 66).332–334 Similar regioisomeric ratios (1.1-2.4:1) favoring reaction at the substituted aryl ring were obtained, independently of the electronic nature of the R group (electron-releasing or electron-withdrawing), thus ruling out an SEAr process for the cyclization. In fact, the experimental results would be better explained by a mechanism where the hydrogen from the phenyl is transferred as a proton in the step deciding the selectivity. B3LYP calculations on the model system PdBr(PH3)(o-(CH2Ph)Ph) estimated the activation barrier for ortho-CsH oxidative addition to Pd at 43.3 kcal mol-1. This originates from the instability of a Pd(IV) intermediate and the too low basicity of bromide to abstract the proton on the phenyl ring. Substitution of Br- by HCO3-, a model for the base used experimentally (CO32-), allowed the location of a TS for proton abstraction assisted by bicarbonate. The activation barrier for the CsH cleavage drops to 23.5 kcal mol-1, an acceptable value for a reaction at 100-135 °C. Introducing three fluorine atoms on the phenyl ring results in an even lower activation barrier (13.2 kcal mol-1), thus confirming the experimental observation that electron-withdrawing substituents accelerate the reaction. An intermolecular pathway, where the base HCO3abstracts the proton without prior coordination of the base to Pd or substitution of bromide, was also probed by the calculations.334 The activation barrier was found to be lower than that in the intramolecular pathway for the phenyl case (17.4 versus 23.5 kcal mol-1). In the case of the fluorinated phenyl ring, the intermolecular pathway is computed to be slightly more difficult than the intramolecular reaction (14.4 versus 13.2 kcal mol-1). Both intermolecular and intramolecular pathways can be operative, and the preference for one over the other may depend on the substrate. For instance, the palladium-catalyzed intramolacular arylation of aryl-bromide can be carried out in the presence of a bidentate phosphine under mild conditions.334 DFT(B3LYP) calculations of the TS for the intermolecular proton abstraction were carried out on model systems for the monodentate and bidentate phosphine complexes (Figure 67). The activation barriers are of similar magnitude (∆Gq ) 17.2 kcal mol-1, 60; ∆Gq ) 16.1 kcal mol-1, 60-F; ∆Gq ) 20.1 kcal mol-1, 61; ∆Gq ) 18.9 kcal mol-1, 61-F) and indicate that the fluorinated ring is preferentially activated (ratio ca. 4:1).

Figure 66. Pd-catalyzed intramolecular arylation with substituted aryl groups.

786 Chemical Reviews, 2010, Vol. 110, No. 2

Figure 67. Models of the monodentate and bidentate phosphine complexes used in the calculations for the study of the competition between intermolecular base-assisted CsH activation reactions.

Figure 68. General reaction for catalytic direct arylation.

Figure 69. Catalytic intermolecular direct arylation of pentafluorobenzene.

These ratios are in qualitatively good agreement with the ratios observed experimentally (ca. 1.7:1).333 Palladium-catalyzed direct arylation of simple arenes (Figure 68) in synthesis of biaryl molecules is a very interesting alternative to classical cross-coupling strategies where aryl organometallic precursors must be prepared beforehand and are sometimes difficult to synthesize or unstable.335 With direct arylation, the two reacting partners are not part of the same molecule. The CsX oxidative addition does not anchor the other reacting site and has thus no influence on the interaction of the CsH bond with the metal. Fagnou and Woo have carried out a combined experimental and computational study of the reaction of pentafluorobenzene with parabromotoluene (Figure 69).336 The catalyst system is assumed to be a monophosphine-Pd(0) complex, and the first reaction is considered to be CsBr oxidative addition to generate the intermediate Pd(PR3)(4-MeC6H4)(Br). In the B3LYP calculations, this intermediate has been modeled by Pd(PH3)(Ph)(Br), and six different mechanisms for the CsH activation of C6F5H have been considered (Figure 70). For mechanism A, no oxidative addition Pd(IV) product (Pd(PH3)(Ph)(C6F5)(Br)(H)) could be located on the PES, and for mechanism B, no adduct corresponding to an electrophilic attack of Pd on C6F5H could be located. Therefore, these two mechanisms are deemed not to be operative. All attempts to locate a TS for intermolecular proton abstraction from a weak adduct of C6F5H to Pd(PH3)(Ph)(Br) as in mechanism C3 failed, because the bicarbonate anion was found to always bind to the Pd center before abstracting the proton. For the mechanism C1, no TS could be located either because it was shown that the reverse

Balcells et al.

reaction is barrierless. The energy of the reaction Pd(PH3)(Ph)(Br)(C6F5H) f Pd(PH3)(Ph)(Br)(C6F5) + HBr is taken as an estimate of the activation barrier along this mechanism. Finally, two mechanisms where CsH cleavage is assisted by coordination of acetate (C2 and C4) have led to the characterization of a proper TS structure on the PES. The main difference between the two mechanisms is the nature of the active species and the calculations indicate that the energy barrier is ∆Eq ) 9.9 kcal mol-1 for C1 and ∆Eq ) 13.3 kcal mol-1 for C4. Given the high temperature used for the reaction, no preference can be attributed to a particular mechanism. However, competition experiments were performed between various fluoroarenes, and the results were compared with the calculations on the same systems. The CsC coupling occurs selectively at the position maximizing the number of ortho-F’s, which has been attributed to the increased acidity of the CsH bond at this position. The agreement between experimental and computational ratios of regioisomers is much better for mechanism C2 than for mechanism C4. Fagnou has coined the name concerted metalation deprotonation (CMD) for this mechanism where the hydrogen is abstracted by the coordinated base simultaneously as the MsC bond is formed. The characteristics of the CMD mechanism have been analyzed through calculations of CsH activation in a broad range of aromatic substrates.37 The calculations show that activation barriers for CsH activation critically depend on the position of the CsH bond in the ring. To elucidate the origin of this effect, an activation strain analysis was performed (Figure 71). The Edist term reflects the energy needed to distort both reactants to the geometry of the TS. For the palladium complex, the energy range spanned is rather narrow (from 15.4 to 20.4 kcal mol-1). In contrast, for the aromatic substrate, the variation is large (from 29.3 to 50.1 kcal mol-1). Thus the reactivity of the various CsH bonds cannot be explained solely with Brönsted acidity arguments. The Eint term reflects the stabilization introduced by the simultaneous Pd · · · C and O · · · H interactions in the TS and varies between -32.2 and -58.0 kcal mol-1. Further studies are nevertheless needed to better understand the factors controlling Edist and Eint and establish predictive rules for the optimization of palladium-catalyzed biaryl synthesis. The CMD mechanism was shown to be operative in the direct arylation of arene CsH bonds in phenyl-pyridine by a NHC-Ru(II) complex.337 DFT calculations by Maseras probed two different pathways. Coordination of pyridine would normally lead to easy ortho-CsH activation through oxidative addition. However, the Ru(Cl)2(IMe) (IMe ) N,N′dimethylimidazol-2-ylidene) fragment does not allow for any σ-CAM transformations, and the Ru(IV) complex resulting from oxidative addition lies at high energy (∆E ) 28.2 kcal mol-1). Coordination of HCO3- to Ru stabilizes the system by 22.9 kcal mol-1, and from this intermediate, CMD is possible with an energy barrier of 13.9 kcal mol-1 and an exothermic reaction energy of -13.7 kcal mol-1. More aryl and heterocyclic CsH bond functionalizations should be possible with catalysts offering an easy orthometalation pathway via metal/base cooperative proton abstraction. The direct arylation reaction has been extended to the intramolecular activation of aliphatic CsH bonds (Figure 72).338 When different aliphatic groups are present that may undergo reaction, high selectivity is observed for reaction at a methyl substituent over a secondary carbon. To further

CsH Bond Activation in Transition Metal Species

Chemical Reviews, 2010, Vol. 110, No. 2 787

Figure 70. Mechanisms considered for the CsH bond activation of benzene by Pd(PH3)(Ph)(Br) with added HCO3-.

understand the parameters leading to this selectivity, B3LYP calculations examined the mechanism of CsH cleavage. Here again a Pd(IV) pathway involving an oxidative addition was ruled out. In the calculations, the base was modeled by AcO- and the phosphine by PMe3. Within the framework of the CMD mechanism, three different TS structures were located corresponding to CsH activation at various positions (CH3, CH3CH2, and CH3CH2, see Figure 72). At the TS, agostic interaction occurs between the σ(CsH) bond and the Pd(II) atom, resulting in significant weakening of the CsH bond. This interaction and the formation of the O · · · HsC bond compensate for the loss of CsH bonding and explain the relatively low activation barriers. The lowest value is obtained for the activation at the methyl group forming the six-membered palladacycle, in agreement with the experimental observations. The CMD mechanism was also shown to operate in the synthesis of benzocyclobutene by palladium-catalyzed CsH activation of methyl groups (Figure 73).339 For the reaction

Figure 71. Scheme for the activation-strain analysis with definition of Edist and Eint and their relation to ∆Eq.

788 Chemical Reviews, 2010, Vol. 110, No. 2

Figure 72. Palladium-catalyzed intramolecular alkane arylation.

Balcells et al.

following section, we describe three groups of reactions where the computational studies help to understand the catalytic nature of the reaction. The first subsection on dehydrogenation of alkanes has been complemented by studies on stoichiometric dehydrogenation of alkanes where computational results help to understand better the crucial step of dehydrogenation.

4.1. Alkane Dehydrogenation Figure 73. Optimized conditions for the synthesis of benzocyclobutenes.

of 2-tBu-bromobenzene, the complete catalytic cycle with Pd(PtBu3) as the model for catalyst was computed at the B3PW91 level with inclusion of the influence of the solvent DMF (PCM single point calculations). The influence of the base was also tested by considering three different cases (acetate, bicarbonate, and carbonate). Of the three different steps (C-Br oxidative addition, CsH activation, and CsC coupling), the intramolecular CsH activation by Pd(II) complexes presents the highest energy barrier (∆Eq ) 4.7 kcal mol-1, C-Br oxidative addition; ∆Eq ) 27.6 kcal mol-1, CsH oxidative addition with CO32-; ∆Eq ) 22.5 kcal mol-1, CsC coupling reaction). Experimentally, the nature of the base is shown to influence drastically the yield of the reaction with quantitative transformation with carbonate, 80% yield with bicarbonate, and only 10% yield with acetate. In the square planar geometry traditionally proposed for CMD transition states at Pd(II), the aryl and the phosphine are cis, thus leading to a cis relative position for the base and the CsH bond to be cleaved. Such a geometry in the TS failed to reproduce the experimental result since the energy barriers were calculated to vary as acetate < bicarbonate < carbonate. The high barrier for CO32- is due to the necessary κ2-κ1 isomerization of the base before actual proton abstraction. In another geometry for the TS, where aryl and phosphine are trans, only κ1 coordination of the base trans to the agostic CsH bond allows proton transfer. In that case, the relative values for the energy barriers, carbonate < bicarbonate < acetate, are in agreement with the relative yields observed experimentally. Moreover, this TS structure also allows proton transfer to the aromatic ring, thus providing an explanation for the formal 1,4 migration observed with some substituted aromatic rings. Heteroatom-assisted CsH activation by carboxylate or carbonate bases are particularly prominent in the development of new catalysts where CsH activation can be exploited in synthesis. The common features of such processes appear to be the simultaneous ambiphilic activation by a Lewis acidic metal center and an intramolecular base. Macgregor and Davies have suggested using the acronym AMLA (ambiphilic metal-ligand activation) for such reactions.38

4. Catalytic Alkane and Arene Functionalization Catalytic reactions based on CsH activation are not yet frequent, and computational studies of the full cycle are even rarer. Very often, computational studies of catalytic reactions have focused on some selected parts of the cycle. This is the case of many reactions involving chelation-assisted CsH activation (section 3.2) or base-assisted CsH activation (section 3.4). Several reactions based on CsH σ-bond metathesis are catalytic and described in section 2.2. In the

Alkane dehydrogenation is highly demanding thermodynamically and kinetically because of the strength of the CsH bond (BDE(CsH) = 105 kcal mol-1 in methane) and also because it is the reverse of a 2π + 2σ addition typically forbidden by symmetry rules. A transition metal fragment can mediate an alternative path to the forbidden direct reaction. The endothermic aspect of the reaction can be dealt with by a sacrificial hydrogen acceptor. Therefore alkane dehydrogenation can be achieved if it becomes an alkane transfer dehydrogenation using an olefin as a sacrificial hydrogen acceptor. Several transition metal complexes have been found to be able to catalyze such reaction. The first catalysts were made by Felkin with Re, Ru, and Ir polyhydrides340,341 and also by Crabtree for Ir polyhydride.342,343 The next generation of catalysts were Rh catalysts.344,345 Jensen and Kaska discovered that iridium pincer complexes (R′PCP)IrH2 where R′PCP ) κ3-2,6-(R′2PCH2)2C6H3Ir were active catalysts.19,346,347 Acceptorless dehydrogenation is a reaction that is even more attractive because it does not require the sacrificial olefin, and the loss of H2 expelled from solution makes the reaction thermodynamically possible. The proof of principal of this reaction (reflux-driven removal of H2 was first shown by Saito348 and by Crabtree,349 and efficiencywasgreatlyincreasedwiththeIrpincercomplexes.40,350 This reaction led to the remarkable discovery of the catalytic alkane metathesis by tandem alkane dehydrogenation/olefin metathesis.42 It should be noted that the reaction in the presence of sacrificial olefin occurs at relatively low temperature (T < 100 °C), while the acceptorless reaction requires higher temperature (T > 150 °C). The catalytic dehydrogenation of an alkane RsH by a metal fragment MLn is mechanistically a simple reaction made of two successive CsH bond cleavages: (1) in the first step, the CsH bond of the alkane RH is oxidatively added to the metal fragment LnM to form the hydrido alkyl complex, LnM(H)(R); (2) in the second step, a β-H elimination from the alkyl complex forms the olefin and leaves an additional hydride on the metal to form LnMH2. LnMH2 has to return to LnM by losing the two hydrogen atoms. Computational studies were carried out on the reaction of the pincer dihydrido Ir(III) complex, (R′PCP)IrH2, with various alkanes to determine how the reaction occurs in presence of sacrificial olefin and in the acceptorless conditions.351–358 The studies focused on the nature of the mechanism, which can be either dissociative or associative with or without the sacrificial olefin (Figure 74). The reaction with sacrificial olefin was studied by Hall et al. with DFT(B3LYP) calculations for the iridium pincer complex with κ3-2,6-(H2PCH2)2C6H3) as the PCP ligand, ethylene as the hydrogen acceptor, and ethane as the substrate.352 The calculations show that the transfer dehydrogenation of ethane by (PCP)IrH2 occurs with two distinct stages. First, (PCP)IrH2 is dehydrogenated by the hydrogen acceptor (coordination of C2H4 cis to the PCP ligand, followed by insertion of ethylene into the IrsH bond and

CsH Bond Activation in Transition Metal Species

Chemical Reviews, 2010, Vol. 110, No. 2 789

Figure 74. (PCP)Ir-catalyzed acceptorless alkane dehydrogenation (left) possible associative (A) and dissociative (D) pathways and (PCP)Ircatalyzed alkane transfer-dehydrogenation (right) possible associative (A) and dissociative (D) pathways.

by the reductive elimination of ethane). This yields the 14electron (PCP)Ir complex that reacts with ethane by way of oxidative addition into the CsH bond, β-H elimination, and dissociation of ethylene cis to the PCP ligand with regeneration of the catalyst. The formation of the unsaturated intermediate (PCP)Ir is an endothermic reaction by 12.7 kcal mol-1 and proceeds by a cis-pathway. The critical steps of the reaction are the hydride transfer to ethylene, ethane oxidative addition, and dissociation of the coordinated ethylene with energy barriers of 14, 11.6, and 23.4 kcal mol-1, respectively. The acceptorless reaction with the same model is strongly energetically disfavored.351,356 Calculations with the anthraphos ligand, which renders the catalyst thermodynamically stable have shown that the dissociative (by way of an Ir(I) intermediate) and the associative pathway (by way of an Ir(V) intermediate), as well as a concerted pathway where H2 leaves as the alkane adds, have similar activation barriers at 298 K.354 Krogh-Jespersen and Goldman have carried out a series of computational studies on the reaction mechanism with a particular emphasis on the role of the temperature in the acceptorless process.353,355,357,358 In their first study,353 they used DFT(B3LYP) calculations to examine the CsH bond cleavage in alkanes (RsH ) CH3-H, C2H5-H, C3H7-H) and arenes (RsH ) C6H5-H) by both (PCP)Ir and (PCP)IrH2. The conclusions are different using the electronic energy barriers (∆Eq, no ZPE corrections) and the Gibbs free energy barriers (∆Gq). Using the electronic energy values, the associative mechanism is preferred over the dissociative mechanism. However the two pathways have comparable Gibbs free energy barriers with a marginal preference for the dissociative pathway. This preference will increase at higher temperature and with both bulkier phosphine and alkane substrates. The dissociative pathway is therefore also preferred for the hydrocarbon/hydrocarbon exchange, which is required in the case of transfer dehydrogenation because the coordination energy of alkanes is very small. The acceptorless reaction occurs only at high temperature. The lowest temperature at which the reaction was found to be possible is 150 °C for the easily dehydrogenated substrate cyclooctane. Combining computational chemistry with classical thermodynamic considerations led Krogh-Jespersen and Goldman to show that the dissociative pathway was significantly preferred at the temperatures used experimentally.355 The dissociative pathway for the reaction of (MePCP)IrH2 with cyclohexane and propane is calculated to have an energy barrier of ca. 36 kcal mol-1; the highest transition state

corresponds to the oxidative addition of the alkane to the 14-electron (MePCP)Ir fragment obtained after loss of H2. This energy barrier transforms into a free energy barrier of 36 kcal mol-1 at standard thermodynamic conditions (PH2 ) 1 atm and T ) 298 K). However, a free energy barrier of only 18 kcal mol-1 for the CsH activation step is calculated for conditions under which the dehydrogenation is thermodynamically possible (T ) 423 K and PH2 ) 10-7 atm with concentration of hydrocarbon CsH bonds equal to those in neat solution). The rate-determining step of the catalytic cycle is now the loss of H2 from (MePCP)IrH2 with an activation energy barrier of 27.2 kcal mol-1. The associative pathway has an energy barrier for the CsH activation step of ∆Eq ) 20.1 kcal mol-1 and for the whole reaction of ∆Eq ) 28.4 kcal mol-1 (loss of H2). The energy barriers translate into Gibbs free energy barriers of ∆Gq ) 32.2 kcal mol-1 for the CsH bond activation step and ∆Gq ) 37.4 kcal mol-1 for subsequent loss of H2 at standard thermodynamic conditions. In the simulation of the experimental conditions, the respective values for the Gibbs free energy barriers for the CsH addition and H2 loss are 31.3 and 36.0 kcal mol-1. The preference for the dissociative pathway is thus nonambiguous. The dissociative pathway is shown to account also for the lack of H for D exchange in cyclohexane-d12. Substitution of the pincer ligand at the para position by a π donor such as methoxy groups gives catalytic activity higher than unsubstituted analogues.358 However, DFT calculations do not provide a clear interpretation of the substituent effects because the methoxy group stabilizes the transition state as much as the resting state. A computational and experimental study of the effect on catalytic activity resulting from systematically varying steric crowding by the substitution of one methyl group for one of the phosphine tert-butyl t groups of the Bu4PCP ligand has been recently published.359 The electronic structure calculations show that a single Mefor-tBu substitution has a favorable energetic effect on the β-H elimination step. However, the stabilizing effect on the overall barrier is shown to be partially offset by the stronger t binding of the 1-alkene to the Me Bu3PCPIr catalyst. The relation between alkane dehydrogenation and alkene H for D exchange was considered computationally by Eisenstein and Crabtree for IrH2(O2CCF3)2(PAr3)2 (Ar ) p-FC6H4).88 It is shown that the H/D scrambling involves the insertion of the olefin to give an alkyl hydride complex that reductively eliminates to lead to a transition state that contains an η3-bound alkane. The relatively large binding

790 Chemical Reviews, 2010, Vol. 110, No. 2

Balcells et al.

Figure 75. Alkane dehydrogenation in a Rh(I) complex via an isolated agostic intermediate.

Figure 76. Intramolecular alkyl phosphine dehydrogenation in cationic rhodium complexes of P(Cyp)3 (Cyp ) cyclopentyl).

energy of the alkane, which is ascribed to the poor electron donating power of O2CCF3, allows the H for D exchange to occur. Alkyl dehydrogenation in a Rh(I) complex via an isolated agostic intermediate was studied by a combined experimental and computational approach by Macgregor and Weller.360 The T-shaped 14-electron rhodium phosphine complex [Rh(PiPr3)3][BArF4] (ArF ) 3,5-(CF3)2C6H3 was shown to contain an unusual β-CsH agostic interaction from the isopropyl phosphine ligand and undergoes intramolecular dehydrogenation (Figure 75). DFT calculations on the cationic system Rh(PH2iPr)(PH3)2+ show that several structures are possible. The most stable is a γ-agostic system, while the β-agostic complex is 5.6 kcal mol-1 higher in energy. CsH activation in the β- and γ-agostic complexes yield three- and four-membered metallacycles, respectively. The latter is more stable. The calculations show that only the metallacyclophosphapropane can achieve full dehydrogenation, as observed experimentally. This shows that only the product of CsH activation at the β-position can lead to productive dehydrogenation. Weller and Macgregor have carried out an experimental and computational study of Rh(nbd)(PCyp3)2 (nbd ) norbornadiene, PCyp3 ) tri(cyclopentylphosphine)), which spontaneously undergoes dehydrogenation of each PCyp3 ligand in CH2Cl2 solution to form an equilibrium mixture of cis- and trans-[Rh(PCyp2(η2-C5H7)]2[BArF4] (Figure 76).361 Furthermore competition experiments using mixed cyclohexyl cyclopentyl phosphine ligand PCy2(Cyp) show that RhCl(nbd)(PCy2(Cyp))2+ undergoes dehydrogenation at the cyclopentyl group. The DFT(BP86) calculations were carried out on the neutral complexes Rh(dppe)PCyp3 and Rh(dppe)PCy3. The CsH activation in the cyclopentylphosphine complex forms directly the rhodium hydride complex with an energy barrier of 9.9 kcal mol-1. The corresponding reaction in the cyclohexylphosphine complex has a slightly higher energy barrier of 13.6 kcal mol-1 and a conformational change in the chairlike cyclohexyl substituent is required to yield the final hydrido product. The lower energy barriers for the CsH activation in the cyclopentyl case are related to the presence of agostic interactions in the reactant. Interestingly, an additional agostic CsH interaction appears at the transition state in the cyclopentyl case (Rh · · · C ) 2.47 Å, CsH ) 1.11 Å), while the corresponding Rh · · · H distance is over 3 Å in the cyclohexyl species. The final dehydrogenation requires β-H transfer and H2 elimination. The potential energy surfaces for the two systems are parallel for this stage of the reaction, but that for cyclohexyl is 5

Figure 77. Computed catalytic cycle (B3LYP) for the insertion of ethylene in the ortho-CsH bond of benzaldehyde. The energies (kcal mol-1) of the various intermediates are expressed relative to the separated reactants, benzaldehyde and Ru(PH3)2(CO), and the energy barriers are given for each elementary step. P stands for PH3.

kcal mol-1 higher in energy than that for the cyclopentyl case. The results are in good agreement with the experimental results of this paper and the known greater propensity for cyclopentylphosphine to dehydrogenate further. This property led to interesting complexes with reversible H2 uptake362 or formation of Ru(0) complexes.363,364

4.2. Hydroarylation of Olefins The Murai reaction is one of the early examples of catalytic functionalization of an aromatic CsH bond.16 The reaction consists of the selective insertion of an olefin into an aromatic CsH bond ortho to an orientating group. Even though Jordan has shown catalytic insertion of propene into the ortho-CsH bond of R-picoline in 1989,281 the reaction was shown to be limited to R-picoline and propene. The Murai reaction, in contrast, shows a larger scope of substrates that are able to insert a great variety of alkenes. A typical catalyst for this reaction is Ru(H)2(CO)(PPh3)3, and the transformation is usually carried out in refluxing toluene. The active species is assumed to be a Ru(0) complex after hydrogen abstraction by the alkene in excess under the catalytic conditions. A computational study at the B3LYP level was carried out by Morokuma to characterize the entire catalytic cycle.365 Two different active species were considered in the model, Ru(CO)(PH3)2 and Ru(CO)(PH3)3. The substrate was benzaldehyde and ethylene was the inserting alkene. The most favorable pathway starts with O-coordination of benzaldehyde trans to the carbonyl group of Ru(CO)(PH3)2 with a stabilization of ∆E ) -25.9 kcal mol-1 (Figure 77). The CsH activation is a two-step process with the formation of an agostic interaction (∆Eq ) 3.3 kcal mol-1) followed by the actual cleavage of the CsH bond (∆Eq ) 2.8 kcal mol-1). These two energy barriers are very low and show the critical influence of the coordination of the carbonyl

CsH Bond Activation in Transition Metal Species

group to ease the activation of the ortho-CsH bond. From the CsH activation product, substitution of the labile phosphine trans to the RusC bond by ethylene affords the required cis-geometry for alkene and H to facilitate the alkene insertion into the Ru-H bond. The latter is endothermic by ∆E ) 9 kcal mol-1 with a low energy barrier (∆Eq ) 11.0 kcal mol-1). The high energy of the insertion product is a result of creating the new RusC bond trans to the already existing RusC bond with the aromatic ring. Loss of the agostic interaction trans to phosphine, associated with the rotation of the ethyl ligand around the RusC bond (∆Eq ) 12.6 kcal mol-1), allows the latter to be cis to the aryl group. This isomer is 1.7 kcal mol-1 more stable than the hydridoalkene complex. Further stabilization is achieved through PH3 coordination at the vacant site. The complex obtained presents the ideal geometry for CsC reductive reaction. Despite the favorable geometrical features, the CsC coupling is the rate-determining step with a calculated energy barrier ∆Eq of 27.0 kcal mol-1, in agreement with the experimental observations. An interesting feature of homogeneous catalysis is to provide synthetic strategies that allow formation of products traditionally difficult to obtain. Periana has shown that bis(acac-O,O)-Ir(III) complexes (acac-O,O ) κ2-O,O-acetylacetonate) are active catalysts in the anti-Markovnikov arylation of olefins with benzene to produce straight-chain alkylbenzenes with higher selectivity than the branched ones.43,366,367 The experimental observation indicated that the catalysis proceeds by arene CsH bond activation by way of a thermally stable homogeneous mononuclear (acacO,O)2Ir(py)(Ph) species (py ) pyridine).43 In a series of papers, Goddard and Periana studied the catalytic cycle of homogeneous Ir(III) catalyzed regioselective arylation of alkenes,200,368,369 even though these reactions have limited use (high temperatures, extended reaction times, limited activity). In the first contribution, the mechanism of hydroarylation of olefins by a homogeneous (acacO,O)2Ir(H2O)(Ph) catalyst is elucidated by B3LYP calculations (Figure 78).200 After dissociation of the water ligand, a vacant site is generated and ethylene coordination is possible. However, coordination of ethylene to the empty coordination site following departure of water would lead to a complex with the phenyl and ethylene ligands in trans coordination, hence preventing the insertion of ethylene in the metal-phenyl bond. The bis-acac complex has to isomerize to bring the two acac ligands in cis coordination, which opens the empty coordination site cis to the phenyl ligand. This isomerization was shown to proceed on the ML5 (acac-O,O)2Ir(Ph) complex with an activation barrier ∆Gq ) 25.9 kcal mol-1 in good agreement with the experimental value of 28.7 kcal mol-1 at 413-473 K. Coordination of ethylene leads to the cis(acac-O,O)2Ir(Ph)(C2H4) intermediate lying slightly above the starting reactants trans-(acac-O,O)2Ir(Ph)(H2O) and ethylene (∆G ) 4.5 kcal mol-1). Insertion of ethylene into the IrsPh bond has an activation barrier of ∆Gq ) 24.8 kcal mol-1 and produces the ethylbenzene complex with coordination of one ortho-CsH bond of benzene. Coordination of a new benzene molecule instead of the ortho-CsH bond is not favored entropically but sets the system for further CsH activation. The latter proceeds along an OHM transfer (see section 2.4) with ∆Gq ) 12.9 kcal mol-1 and ∆G ) 0.2 kcal mol-1. Finally, substitution of the η2-(CsH) bond of

Chemical Reviews, 2010, Vol. 110, No. 2 791

Figure 78. Computed catalytic cycle for the formation of ethylbenzene by insertion of ethylene into the CsH bond of benzene in presence of (acac-O,O)2Ir(Ph)(H2O). Gibbs free energies (kcal mol-1) are expressed relative to ethylene and the water complex. The activation barriers are given for each elementary step.

the ethylbenzene produced by a new ethylene molecule closes the catalytic cycle. The authors find that the rate-determining step is olefin insertion into the IrsPh bond. Calculations of the transition states of olefin insertion for various alkenes (propene, styrene, and isobutene) probed at the regioselectivity between 1,2 and 2,1 insertion. The experimental ratios between the branched and the linear alkylbenzene allow the determination of ∆∆Gq for the rate-determining step. These experimental values can be directly compared with the calculated values from the energies of the two transition states for insertion. The agreement is excellent, and the preference for the linear isomer is shown to be preferentially due to the electrondonating character of the substituent of the alkene. The steric influence of the substituent has a lower influence to prevent insertion leading to a branched isomer. From the ethylbenzene product of ethylene insertion (Figure 78), a β-hydride elimination is possible generating a hydrido-styrene complex. This process is calculated to be easier than benzene coordination because the β-H elimination is an intramolecular reaction. This pathway constitutes a potential deactivation route. However, the dissociation energy of styrene is computed to be larger than the activation barrier for insertion of styrene into IrsH (the reverse of β-elimination). Therefore, if formed, the hydrido-styrene would preferentially revert to the ethylbenzene, and the catalytic cycle could eventually be closed. Inhibition of the catalyst by ethylene was also studied computationally. Again, from the ethylbenzene product of

792 Chemical Reviews, 2010, Vol. 110, No. 2

the ethylene insertion into the IrsPh bond, coordination of ethylene at the vacant site leads to an adduct more stable than benzene coordination. The ethylene complex can follow two different paths, insertion in the IrsC bond to give a butylbenzene intermediate or CsH activation through an OHM process to release the product (ethylbenzene) and a vinyl complex. These two routes are associated with transition states with similar Gibbs free energies. One pathway is productive, while the other leads to multi-insertion products. Calculations indicate that both are likely to occur. This explains the experimental results of diminution of the TOF upon increase of the ethylene/benzene ratio. A detailed analysis of the CsH bond activation of benzene was carried out experimentally and computationally on (acacO,O)2Ir(R)(py) (py ) pyridine; R ) CH3, C2H5, Ph, CH2CHPh).368 Pyridine dissociation easily generates trans(acac-O,O)2Ir(R), and the CsH activation reaction necessitates further isomerization to produce the cis-acac intermediate. This reaction constitutes the bulk of the barrier for the CsH activation. Overall the CsH activation reaction has been shown to proceed in four key steps: (i) a preequilibrium loss of pyridine, (ii) a rate-determining unimolecular isomerization of the trans-(acac-O,O)2Ir(R) to the cis(acac-O,O)2Ir(R) complex, (iii) coordination of benzene to the cis-isomer, and (iv) rapid CsH cleavage. With the ultimate goal of designing more active catalysts for the hydroarylation of unactivated olefins, Periana synthesized (trop-O,O)2Ir(Ph)(py) (trop-O,O ) κ2-O,O-troponolate).369 The catalytic activity of this new system is comparable to that of (acac-O,O)2Ir(Ph)(py). The computed reaction pathway for the transformation was very similar to that computed with the acetylacetonate complex. In an interesting variation, Periana showed that (acacO,O)2Ir(py)(vy) (vy ) HCdCH2) is a catalyst for the hydrovinylation of olefins.370 The products of the reaction are 1-butene and cis- and trans-2-butene in a 1:2:1 ratio. Calculations by Goddard showed that a mechanism similar to that for the hydroarylation of olefin is operative.371 After dissociation of pyridine and trans-cis isomerization, a coordinated ethylene inserts into the Irsvy bond to generate an IrsCH2sCH2sCHdCH2 complex. The latter isomerizes to an Ir(η3-allyl) complex through a series of β-hydride transfer reactions. This η3-allyl intermediate is the most stable species and constitutes the resting state of the catalyst. The η3-η1 isomerization of the allyl opens a coordination site for ethylene followed by CsH activation through an OHM process, which generates the product 2-butene and reforms the vinyl complex. The CsH activation is computed to be rate-determining with ∆Hq ) 31.6 kcal mol-1 and ∆Gq ) 32.1 kcal mol-1. Formation of 1-butene is not favorable from the η3-allyl complex, and the experimental observation of 1-butene is explained by isomerization of 2-butene after release of the latter from Ir. Another system, TpRu(CO)(NCMe)(Ph), was shown by Gunnoe to catalyze the addition of arene to ethylene and propene.44 The catalyst system is mildly selective for the linear alkylbenzene with an observed ratio of 1.6:1 in favor of propylbenzene with respect to isopropylbenzene. The catalytic cycle for this transformation has been determined computationally by Goddard (Figure 79).201 The calculations were carried out using the experimental system and ethylene. The reaction starts with dissociation of acetonitrile and coordination of ethylene. The Tp ligand imposes a fac geometry for the three other ligands, which drives the phenyl

Balcells et al.

and ethylene ligands in cis coordination. The next step is ethylene insertion to produce an ethylbenzene, which interacts weakly with the metal by the benzene ring. This intramolecular benzene coordination is easily converted into an intermolecular coordination of an additional benzene molecule. CsH activation directly transfers the hydrogen atom from benzene to the ethyl group, generating the product without explicit formation of a Ru(IV) intermediate. The TS of CsH cleavage features the short Ru · · · H distance typical of OHM transition states (see section 2.4). The mechanism calculated for the present Ru(II) system has been contrasted with the mechanism calculated for Ir(III).201 In both cases, the rate-determining step is ethylene insertion into MsPh. For this step, the activation barrier for Ru is 3.8 kcal mol-1 lower than that for Ir, thus rationalizing the rate enhancement of 200 at 90 °C observed experimentally between the Ru and Ir catalysts. The CsH bond activation step, on the other hand, is higher in energy in the Ru system. The opposite trends in the two key steps will influence the outcome and yield of the catalytic hydroarylation. Improving CsH activation would most likely impede insertion and thus overall rate, while improving insertion and impeding CsH activation would most likely lead to faster rates but extensive polymerization. A detailed mechanistic analysis of hydroarylation catalysts has been carried out by Periana and Goddard for a large number of metal fragments.372 There are two key steps in the process: (i) insertion of ethylene into the MsPh and (ii) CsH activation of an unactivated benzene. These two key steps were computed for TpRu(CO)(Ph) and cis-(acacO,O)2Ir(Ph), the two complexes shown to be active catalysts, but also for Rh, Pd, Os, and Pt with similar ligand sets. From the extended set of data, it appears that the activation barrier for the two key steps are inversely correlated, complicating optimization of the overall process. The accessibility of the Mn+2 oxidation state is at the heart of this behavior. The CsH activation, as an OHM transfer, develops a Mn+2 oxidized character in the TS, which makes it easier for more easily oxidized metal (Ir(III) vs Ru(II)). An easily energetically accessible Mn+2 state implies that a low-lying unoccupied d orbital is available. The metal is acidic, and coordination of an olefin is stronger, thus making the olefin insertion more difficult through a ground-state stabilization effect. A long-standing collaboration between the experimental group of Gunnoe and the computational group of Cundari led to a detailled study of the catalytic properties of the TpRu system.202,373–378 An account of this collaboration has been recently published.379 The first calculations were carried out on the model (Tab)Ru(CO)(Ph) (Tab ) HB(sNdNH)3) and ethylene.373 The computational study particularly addressed the nature of the CsH activation step and showed that no Ru(IV) intermediate was implied. The complex TpRu(CO)(NCMe)(Me) was shown to initiate CsH bond activation at the 2-position of furan and thiophene to produce TpRu(CO)(NCMe)(Ar) (Ar ) 2-furyl, 2-thienyl).374 The 2-furyl complex serves as a catalyst for the formation of 2-ethylfuran from ethylene and furan. DFT calculations specifically studied the nature of the CsH activation process. Substituting CO with PMe3 in TpRu(CO)(NCMe)(Ph) gave a system that exhibited lower catalytic activity in the hydrophenylation of ethylene.376 For TpRu(PMe3)(NCMe)(Ph), CsH activation of ethylene to ultimately

CsH Bond Activation in Transition Metal Species

Chemical Reviews, 2010, Vol. 110, No. 2 793

Figure 79. Computed catalytic cycle for the formation of ethylbenzene by insertion of ethylene into the CsH bond of benzene in the presence of TpRu(L)(CNMe)(Ph) (L ) CO, PMe3).

produce TpRu(PMe3)(η3-C4H7) is found to kinetically competewithcatalyticethylenehydrophenylation.TheDFT(B3LYP) calculations were carried out with the experimental ligands Tp, CO, and PMe3. The entire catalytic cycle was computed for TpRu(L)Ph (L ) CO and PMe3), and in both cases, the highest point on the PES is the TS for CsH activation of benzene generating the ethylbenzene product. The lower activity in the case of L ) PMe3 is due to easy CsH activation of ethylene by TpRu(PMe3)Ph to form the vinyl complex TpRu(PMe3)(vy). The latter converts to an η3-allyl complex in the presence of excess ethylene. The calculations show that ethylene insertion into RusPh is much easier than ethylene CsH activation for L ) CO (∆∆Gq ) 8.6 kcal mol-1), whereas the difference between the two pathways is less pronounced for L ) PMe3 (∆∆Gq ) 3.1 kcal mol-1 in favor of ethylene insertion). Combined experimental and computational study of hydroarylation of olefins by TpRu{P(pyr)3}(NCMe)Ph (pyr ) N-pyrrolyl) tested the impact of sterics on the catalytic activity.377 The steric bulk of the tris-N-pyrrolyl phosphine ligand inhibits coordination of ethylene and thus prevents CsC bond formation, rendering the catalyst inefficient. In an effort to more precisely delineate the effect of the ancillary ligand L on Ru-mediated olefin hydroarylation activity, Gunnoe prepared and studied the reactivity of TpRu{P(OCH2)3CEt}(NCMe)Ph (P(OCH2)3CEt ) 4-ethyl2,6,7-trioxa-1-phosphabicyclo[2.2.2]octane). The reduced cone angle (101°) of the P(OCH2)3CEt ligand was anticipated to allow olefin coordination, while the moderate π-acidity

of the phosphite was expected to potentially bias the kinetics toward olefin insertion over olefin CsH activation as observed with CO and contrary to the situation with PMe3. A maximum activity of 10 TONs was achieved at 90 °C and 10 psi of ethylene. As in the other cases, the ratedetermining step for the entire catalytic cycle to form the ethylbenzene product is computed to be the CsH bond activation of benzene. However, a crucial step for the deactivation of the catalyst is the competition between ethylene insertion into RusPh and CsH activation of ethylene to produce a vinyl complex that eventually evolves to an η3-allyl complex. In the present case, the ∆∆Gq value of 7.4 kcal mol-1 is intermediate between that for CO (8.6 kcal mol-1) and that for PMe3 (3.1 kcal mol-1), in agreement with a catalytic activity for the phosphite complex intermediate between that of the carbonyl and that of the phosphine complexes.

4.3. Borylation of Alkanes Metal-catalyzed borylation of arenes380 and alkanes46 is a powerful way to functionalize a CsH bond in a selective manner.45 For instance, with Cp*W(CO)3Bcat′ (cat′ ) 1,2O2C6H2-3,5-(CH3)2), the alkane functionalization resulted in remarkably high yields (Figure 80). The mechanism for the borylation of alkanes was first addressed by Wan et al. using BH3 as a model for H-Bpin (pin ) pinacolate) used in the experiments. B3LYP, B3PW91, and MP2 calculations of the reaction of CpRhH(BH2) with

794 Chemical Reviews, 2010, Vol. 110, No. 2

Balcells et al.

Figure 80. Selective functionalization of alkanes by transition metal boryl complexes.

Figure 82. Reaction pathway for the rhodium-phosphinecatalyzed borylation of CsH bonds with pinacolborane.

Figure 81. Proposed general reaction pathway for CpM(CO)m(B(OCH2)2) (M ) Fe, m ) 2; M ) W, m ) 3) reacting with CH4.

CH4 were carried out.381 The formation of the reactive CpRhH(BH2) species by reaction of CpRhH2 with BH3 was also considered, and it was shown that eliminating H2 from CpRh(H)3BH2 is a high-energy process. However, the unsaturated CpRhH(BH2) forms first a complex with CH4 before cleaving the CsH bond. The BsC bond is formed afterward from the isomer where the CH3 and BH2 are in a cis position. The authors mentioned a significant influence of the method of calculation. Both the DFT(B3PW91) and MP2 methods propose a preference for an oxidative addition/ reductive elimination process over a σ-CAM transformation, but the B3LYP method gives different results for no clear reasons. The reaction pathway was studied by Hall using a model for the boryl ligand closer to the experimental systems.382 DFT(B3LYP) calculations on the reaction of CpM(CO)m(B(OCH2)2) (M ) Fe, m ) 2; M ) W, m ) 3) with CH4 were carried out (Figure 81). After formation of a CH4 complex (∆H ) -6.1 kcal mol-1, Fe; ∆H ) -3.7 kcal mol-1, W), the CsH activation occurs by σ-bond methatesis (∆Hq ) 1.3 kcal mol-1, Fe; ∆Hq ) 4.3 kcal mol-1, W; relative to separated reactants) to form a borane σ-complex as an intermediate (∆H ) -5.4 kcal mol-1, Fe; ∆H ) -3.1 kcal mol-1, W). This borane σ-complex rotates from a conformer where H is closer to the CH3 group to one where the B(OR)2 group is closer (cis) to the CH3 group. The next step is a transfer of the B(OR)2 group toward CH3 through a four-center transition state with B(OR)2 at the β-position of the ring. The overall reaction, which is described as a boron-assisted σ-bond metathesis is a nice example of a σ-CAM mechanism where the facile rotation of the borane σ-complex (energy barriers of 6.2 kcal mol-1 for Fe and 4 kcal mol-1 for W) is indispensable for the formation of the BsC bond. The analysis of the electron density shows that the electropositive nature of the boron causes the electron pair in the MsB bond to attack the H; this attack is supported by (i) transfer of the electron pair from RsH to the new MsC bond and (ii) back-donation of electrons on the metal to the boron p orbital.

This work was followed by a combined experimental/ computational study on the rhodium boryl complexes where monoboryl and bisboryl complexes were compared.383 The DFT(B3LYP) studies show that the ancillary boryl ligand on the metal tunes the shape of the potential energy surface without changing the global picture. In all cases, the CsH bond cleavage process occurs by a boron-assisted σ-bond metathesis mechanism to generate a borane complex that isomerizes, if necessary, to place the alkyl group cis to the boryl group. This complex with cis boryl and alkyl groups undergoes BsC bond formation by a second σ-bond metathesis to generate the product. The calculations show that the presence of an ancillary boryl in place of a hydride lowers the energy of all intermediates and transition states but notably that of the final product. This contributes to drive thermodynamically the reaction to completion. It is also shown that the substituents on the boron influence the potential energy surface and that the experimental OR substituent should not be replaced by hydrogen. Methane and benzene borylation by CpFe(CO)(BO2C2H2) and CpW(CO)2(BO2C2H2) were compared with B3LYP calculations.384 The calculations show that the CsH bond cleavage and BsC bond formation occurs in a concerted manner in a one-step mechanism for the Fe complex because Fe does not favor a high oxidation state. In contrast, a twostep reaction with intermediate formation of a hydrido alkyl boryl complex with a W(IV) oxidation state is observed in the case of W. The reaction with benzene is found to have significantly lower barrier with the Fe complex (∆Eq ) 20.3 kcal mol-1, CH4; ∆Eq ) 11.7 kcal mol-1, benzene) in agreement with experiment. This is attributed to a significantly stabilizing interaction between the boron p orbital of the boryl group and the π orbitals of benzene at the transition state. These interactions are negligible for the tungsten complex because the functionalization occurs in a two-step process. Additional calculations on CpRu(CO)(BO2C2H2) predict that the reactivity difference toward alkanes and arenes should be smaller than that of the iron-boryl complexes. The reaction mechanism of the rhodium-phosphine catalyzed borylation of methyl-substituted arenes using pinacolborane(HBin)hasbeeninvestigatedusingDFT(B3PW91) calculations with additional single point calculations at the CCSD(T) level (Figure 82).385 Factors affecting selectivity for benzylic vs aromatic CsH bond activation were

CsH Bond Activation in Transition Metal Species

examined. The study compares the pathway in which RhH(PMe3)2, model for RhH(PiPr3)2, reacts first with toluene and next with H-Bpin with the pathway where the two substrates react in the reverse order. If the reaction starts with the formation of the boryl derivative, the CsH addition to toluene becomes very unfavorable (highest TS on the PES, 31.8 kcal mol-1 above reactants) due to the exceedingly large trans influence of the boryl group.386 This is especially important in this reaction where the reactive metal fragment has a mer-type coordination at Rh, which maximizes trans influence. It is interesting to note that this is not the case with the cyclopentadienyl derivatives described previously, which have a fac-type geometry where boryl is never trans to the incoming ligand. The preference for the benzylic regioselectivity observed experimentally is attributed to the formation of a stable η3-benzyl intermediate. The potential energy surfaces for the reactions at the benzylic and aromatic CH differ only slightly (difference of 2.3 kcal mol-1 between the highest points of the two pathways), but the trend is predicted correctly. It agrees with the 4:1 product ratio of PhCH2Bpin and Me-C6H4Bpin in the borylation of toluene. The catalytic hydroborylation of arenes was studied with DFT calculations by Sakaki et al.387 Following the report of hydroborylation of benzene by Smith et al.,380 who used CpIr(H)(Bpin)(PMe3) (pin ) pinacolato (sOCMe2s CMe2Os)), [IrCl(cod)]2/phosphine, and [Ir(ind)(cod)]/phosphine (ind ) indenyl) as catalysts, Hartwig et al. used a mixture of bis(pinacolato)diboron with [IrCl(cod)]2/bpy (bpy ) 2,2′-bipyridine or 4,4′-di-tert-butyl-2,2′-bypridine) as precatalyst.388,389 The reaction forms C6H5-Bpin and releases dihydrogen. Calculations were carried out at the B3LYP level using a diimine model of the bipyridine ligand for exploratory calculations and the full ligand for the rest of the study. The pinacolate group is modeled by a bis(ethylene glycol) (Beg) group. To model a reaction in solution, the authors consider that it is better not to include the translational and rotational contributions to the entropy. The key steps of the reaction are the oxidative addition of the CsH bond of benzene to the Ir complex, initiated by π-coordination of the arene, and the reductive elimination of the phenylborane. One interesting aspect of this study is that benzene as well as diborane can be oxidized by the various Ir complexes and intermediates. However diborane is more easily oxidized than benzene, which rationalizes the excess of diborane used in the experiment. Therefore, diborane but not benzene reacts with the Ir(I) precatalyst, which leads to the proposal of IrIII(Beg)3(bpy) as the reactive form of the catalyst. This trisboryl Ir(III) complex promotes the oxidative addition of the CsH bond of benzene, yielding the IrV(Beg)3(H)(Ph)(bpy) complex. The reductive elimination of phenylborane takes place from this species with the concomitant formation of the IrIII(H)(Beg)2(bpy) complex. The CsH oxidative addition has an activation barrier of 24.2 kcal mol-1, and the reductive elimination of phenyl borane has a low activation barrier of 4.9 kcal mol-1. The complex IrIII(H)(Beg)2(bpy) reacts preferably with Beg-Beg with a low activation barrier of 8 kcal mol-1 to form an IrV(Beg)3(bpy) complex, which eliminates H-Beg with a very low activation barrier of 2.6 kcal mol-1 and regenerates the catalyst, IrIII(Beg)3(bpy). After consumption of the diborane, the catalyst is regenerated by reaction of IrIII(H)(Beg)2(bpy) with H-Beg while H2 is eliminated. The charge analysis shows that the Beg group stabilizes the IrV oxidation state, in agreement with the large electron-donating effect of the

Chemical Reviews, 2010, Vol. 110, No. 2 795

Figure 83. The Shilov reaction.

boryl group.386 This study is confirmed by the reaction of square based diphosphine trisboryl IrIII complex with the C(sp2)sH bond of arene and thiophene derivatives.390

5. CsH Oxidation 5.1. Shilov Chemistry: Methane Oxidation Garnett and Hodges originally discovered that [Pt(Cl)4]2promotes H/D isotope exchange in arenes.391 Following this work, Shilov showed that the inert CsH bonds of several alkanes were activated by [Pt(Cl)4]2- in acidic aqueous conditions (Figure 83).1,392 This transformation, known as the Shilov reaction and developed in the period 1970-1985, revolutionized the field of alkane functionalization because, at that time, it was believed that inert CsH bonds could be only activated by metal surfaces, strong oxidants, or radical reagents. In addition, the Shilov reaction obeyed the following selectivity pattern: primary CsH > secondary CsH > tertiary CsH. This pattern was unexpected and opposite to that given by electrophilic and radical reagents. The selectivity suggested that the CsH bond is cleaved through oxidative addition, but this was felt at the time to be unlikely since it would involve the oxidation of Pt(II) to Pt(IV). Another puzzling feature of the Shilov reaction is the fact that it works under aqueous conditions, in which water competes strongly with the alkane substrates for coordination to the metal. Several theoretical studies have focused on the reaction mechanism of the Shilov reaction in order to rationalize its puzzling features and facilitate its further development and optimization. Despite all experimental efforts, the exact identity of the active species is not fully clear yet, and the theoretical studies are indeed based on different models of these species. These studies have shown that the nature of the ligands and their spatial arrangement around the metal center have a strong influence on the reaction mechanism. The review by Lersch and Tilset summarizes the mechanistic studies on CsH activation by platinum complexes up to 2005.57 The Shilov reaction was theoretically studied by Siegbahn and Crabtree, using DFT and ab initio methods.393 The CsH activation of methane by trans-[Pt(Cl)2(H2O)2] was considered as a model reaction (Figure 84). The mechanism starts with the substitution of one aqua ligand by methane, which has an energy cost of 10 kcal mol-1. The next step has been described by Crabtree as a σ-bond metathesis in which one H of methane is transferred to a Cl ligand yielding HCl and a PtsCH3 bond. However, it can be alternatively described as an ambiphilic metal-ligand activation (AMLA) since the lone pair of the Cl ligand assists the deprotonation of methane. This reaction has an energy barrier of 16.5 kcal mol-1. Considering the initial substitution step, the overall activation energy of ca. 27 kcal mol-1 is in remarkable agreement with the experimental value of 28 kcal mol-1. In the final step, an intramolecular proton transfer from HCl to one of the water molecules stabilizes the system by ca. 12 kcal mol-1. The σ-bond metathesis or AMLA mechanism avoids the formation of Pt(IV). The participation of water

796 Chemical Reviews, 2010, Vol. 110, No. 2

Balcells et al.

Figure 84. The mechanism proposed by Siegbahn and Crabtree for the Shilov reaction.

Figure 86. Schematic diagram with the key molecular orbitals of the trans XsPts(CH) fragment. Figure 85. The methane uptake/oxidative addition mechanism proposed by Zhu and Ziegler for the Shilov reaction.

in the first and third steps highlights the critical role played by the solvent. The origin of the unusual selectivity, primary CsH > secondary CsH, was traced to the higher polarity and strength of the Pts(n-alkyl) bond with respect to the Pts(isoalkyl) bond. An alternative oxidative addition mechanism for CsH activation is also possible and may compete with σ-bond metathesis. The predominant transinfluence of the PtsR bond over the other ligands prevents the formation of a vacant site cis to the alkyl R ligand.394 When R contains more than one C atom, this prevents β-elimination, which would erode the selectivity of the reaction by yielding the corresponding alkene. The mechanism of the Shilov reaction has been also explored by Zhu and Ziegler, in a series of theoretical studies.395–398 In the first study, the isotope H/D exchange in methane promoted by [Pt(Cl)4]2- was investigated using a DFT method.395 This reaction involves the following four steps: (1) methane uptake, in which one of the ligands bound to Pt is replaced by methane, (2) CsH oxidative addition, yielding a Pt(IV) hydride intermediate, (3) H/D exchange, and (4) extrusion of the deuterated product, CH3D, by reductive elimination. The reactions leading to CsH activation (Figure 85) were analyzed in detail, and the calculations indicated that methane uptake is the rate-determining step. However, the mechanistic nature of this step, which can be either associative or dissociative, could not be established due to the limitations of the theoretical method in predicting accurate free energies. In order to solve this problem, an alternative kinetic approach was used by deriving a H/D exchange rate expression for each mechanism. When [Pt(Cl)4]2- is used as catalyst in acidic aqueous solutions, a mixture of [Pt(Cl)4]2-, [Pt(Cl)3(H2O)]-, and [Pt(Cl)2(H2O)2] is formed by Cl-/H2O ligand substitution. Each one of these species contributes to the overall rate of the reaction, depending on their concentration and energy barrier for methane uptake. For the dissociative mechanism, the largest contribution to the overall rate comes from [Pt(Cl)4]2-, whereas for the associative mechanism, it comes from

[Pt(Cl)3(H2O)]-. For both mechanisms, the contribution of [Pt(Cl)2(H2O)2] is negligible due to the extremely low concentration of this species. [Pt(Cl)2(H2O)2] becomes the dominant active species only when it is directly used as catalyst. The calculations support the associative mechanism because it gives the best agreement between the experimental and theoretical rates. In a subsequent study by the same authors, the activation of methane by 1:2 mixtures of [Pt(Cl)2(H2O)2] and X (X ) F-, Cl-, Br-, I-, NO2-, and CN-) was investigated, by considering the possible Cl-/X and H2O/X ligand substitutions in aqueous solution.396 In all cases, methane uptake is the rate-determining step, rather than the cleavage of the CsH bond. The ligand pattern around the square planar Pt center of the active species depends on the nature of X. For the weak coordinating X ligands, Cl-, Br-, and I-, the active species are cis-[Pt(Cl)(H2O)(X)2]- and trans[Pt(Cl)2(X)(H2O)]-, whereas the dominant [Pt(Cl)2(X)2]2complex has no noticeable activity. For the strong coordinating X ligands, F-, NO2-, and CN-, a complex mixture of Pt species, including [Pt(Cl)2(H2O)2], [Pt(Cl)2(X)(H2O)]-, [Pt(Cl)(X)2(H2O)]-, [Pt(Cl)(X)3]2-, and [Pt(Cl)4]2-, is formed. Of all these complexes, only cis-[Pt(Cl)2(H2O)2] and cis[Pt(Cl)2(X)(H2O)]- are efficient catalysts. This study showed that catalytic activity requires the presence of an aqua ligand acting as leaving group in the rate-determining methane uptake. In the absence of aqua ligands, the X ligands, which are much poorer leaving groups than water, yield high energy barriers. The calculations by Zhu and Ziegler revealed that the cis/ trans isomerism in the active Pt aqua species plays a key role, because methane uptake is facilitated by strong labilizing ligands trans to the H2O ligand. Nevertheless, as the trans-labilizing strength of the ligand increases, the energy barrier associated with CsH cleavage becomes higher. This undesired effect was found in a study of the CsH activation of methane by trans-[Pt(Cl)2(X)(CH4)]-, in which the X ligand (F-, Cl-, Br-, I-, NO2-, or CN-) lies trans to CH4.397 The bonding in the XsPtsCH4 fragment involves both donation from the CH4 σCH orbital to the PtsX 2a1 orbital and back-donation from the PtsX 1b1 orbital to the CH4 σCH* orbital (Figure 86). As the electronegativity of X

CsH Bond Activation in Transition Metal Species

Figure 87. Reaction mechanism for the catalytic oxidation of methane proposed by Periana. The chelating N,N ligand is bipyrimidine and X ) Cl or OSO3H.

decreases, F- > Cl- > Br- > I- > NO2- > CN-, the contribution of X into the 2a1 and 1b1 orbitals increases, whereas the contribution of Pt to the same orbitals becomes smaller. As a consequence, the 〈σCH|2a1〉 and 〈σCH*|1b1〉 overlaps decrease, and the PtsCH4 bond is weakened. This effect is stronger in the transition state than in the reactant adduct, and therefore, the CsH cleavage energy barrier increases following the order F- < Cl- < Br- < I- < NO2< CN-. These calculations showed that the trans-labilizing effect of the X ligand needs to be finely tuned to achieve the optimal balance between the energy costs associated with CsH cleavage and H2O/CH4 ligand substitution. This principle was further confirmed in a recent study by considering many other ligands, including phosphines, amines, and imines.398 The Shilov chemistry was further developed by Periana, who reported the Catalytica process for the oxidation of methane to methanol.4,399–401 In this process, methane is catalytically oxidized by a Pt(II) complex, [Pt(Cl)2(bpym)] (bpym ) bipyrimidine), under moderate conditions using concentrated sufuric acid as solvent and oxidant.49 The product of the reaction is a mixture of methyl bisulfate (CH3OSO3H), protonated methanol, and methanol, depending on the concentration of sulfuric acid. Both the methyl bisulfate and protonated methanol subproducts are easily converted to methanol by hydrolysis. Periana proposed a mechanism for this reaction (Figure 87), which consists of four steps: (1) dissociation of one of the X ligands, which can be either Cl- or OSO3H-, yielding a T-shaped Pt(II) complex, [Pt(X)(bpym)]+, (2) formation of a Pt(II)sCH3 intermediate, [Pt(X)(CH3)(bpym)], by CsH activation, (3) oxidation by SO3 to a Pt(IV) octahedral complex, [Pt(X)2(CH3)(OSO3H)(bpym)], and (4) functionalization by reductive elimination of CH3OSO3H, in which the catalyst is recovered. A theoretical study by Hush using cisplatin, [Pt(Cl)2(NH3)2], as model catalyst, showed that this reaction mechanism is thermodynamically feasible.402 The calculations suggested that a H2SO4-solvated form of the postulated T-shaped intermediate, [Pt(OSO3H)(OSO3H2)(NH3)2]+, is likely to be the active catalytic species. In this complex, the anionic OSO3H- ligand is bound to Pt through the deprotonated O, whereas the neutral OSO3H2 ligand is bound to Pt through one oxygen of the SdO groups. An alternative mechanism, involving the formation of a Pt-H species by oxidative addition of the CsH bond, is excluded.

Chemical Reviews, 2010, Vol. 110, No. 2 797

Figure 88. The (1) σ-bond metathesis and (2) oxidative addition mechanisms for the CsH activation of methane by cis[Pt(NH3)2(Cl)2].

These studies were further extended by the same authors.403 Thermodynamic and kinetic aspects of the mechanism of methane oxidation by cis- and transplatin were studied. The reaction is initiated by the substitution of an NH3 ligand by CH4. This step, with an energy barrier of ca. 34 kcal mol-1 for the cis isomer and ca. 44 kcal mol-1 for the trans isomer, is effectively rate-determining. The large energy needed for the NH3/CH4 substitution is associated with the high strength of the PtsNH3 bond. The following steps, in which methane is activated, involve similar but slightly lower energy barriers. The CsH bond of methane can be cleaved by either σ-bond metathesis or oxidative addition (Figure 88). The metathesis mechanism (or AMLA), which preserves the Pt(II) oxidation state, involves a proton transfer from methane to a Cl- ligand. The oxidative addition mechanism yields an octahedral Pt(IV) hydride intermediate. Following this step, the reaction is completed by reductive elimination of NH4+. The PtsCH3 and PtsH bonds obtained by oxidative addition are always cis to each other due to trans influence. In addition, the nature of the mechanism depends on the cis/trans isomerism of the catalyst. For cisplatin, the metathesis and oxidative addition mechanisms have similar energy profiles and may compete. In contrast, for transplatin, the oxidative addition mechanism is favored over the σ-bond metathesis mechanism. A theoretical study by Gilbert and Ziegler, with a DFT method including solvation, suggested that the active species in the Catalytica process is either [Pt(bipym)(Cl)]+ or [Pt(bipym)(OSO3H)]+.404 The inclusion of the full bipym ligand in the model revealed that the system does not undergo the dissociation of the Pt-N bond, due to the bidentate nature of the real ligand. In addition, the axial attack of methane to the square planar [Pt(bipym)(Cl)2] catalyst was excluded. Interestingly, the CsH activation mechanism depends on the nature of the catalyst: it is an oxidative addition for [Pt(bipym)(Cl)]+ but a σ-bond metathesis for [Pt(bipym)(OSO3H)]+. The protonation of the bipym ligand is unfavorable but promotes σ-bond metathesis by lowering the associated energy barriers. The calculations showed that the need for high temperatures, 180-220 °C, is due to the high energy needed to substitute a Cl- or a OSO3H- ligand by methane. The reaction mechanism of methane oxidation catalyzed by cis-[Pt(NH3)2(Cl)2] and [Pt(bpym)(Cl)2] was theoretically investigated by Periana and Goddard.405 The potential substitution of a Cl- ligand by OSO3H- was considered for each step of the original mechanistic proposal (Figure 87).

798 Chemical Reviews, 2010, Vol. 110, No. 2

The initial coordination and CsH activation of methane involve [Pt(L2)(Cl)2] (L2 ) bpym or (NH3)2), in which one of the Cl- ligands is replaced by CH3. The Pt(IV) octahedral intermediate formed in the oxidation step has X ) OSO3Hligands in the trans apical positions. The final functionalization step yields [Pt(L2)(Cl)(OSO3H)], which by OSO3H-/ CH4 ligand exchange closes the catalytic cycle. In the CsH activation step, the σ-bond metathesis mechanism, referred to as electrophilic substitution by Periana and Goddard, is preferred for the [Pt(bpym)(Cl)2] catalyst. In contrast, when the catalyst is cis-[Pt(NH3)2(Cl)2], the most favored mechanism is oxidative addition. The oxidation of Pt(II) to Pt(IV) by SO3 (Figure 87) has been studied in detail theoretically. DFT calculations by Hristov and Ziegler showed that this reaction involves the coordination of SO3 to the PtsCH3 intermediate.406 After protonation of the SO3 ligand, Pt(II) is oxidized to Pt(IV) by S-OH bond cleavage in the resulting SO3H ligand, which leads to a cis-Pt(SO2)(OH) species. Final dissociation of SO2 and OH-/OSO3H- ligand exchange yields the cisPt(OSO3H)(CH3) intermediate, which leads to the final reaction product by reductive elimination. Other calculations by Goddard405,407 and Musgrave408 showed that the oxidation step is more favorable for [Pt(NH3)2(Cl)2] than for [Pt(bpym)(Cl)2], thus suggesting that the former complex has a higher catalytic activity. Nevertheless, [Pt(bpym)(Cl)2] was found to be the most robust catalyst, in good agreement with the experiments, because it is more resistant to the strong acidic media associated with the concentrated sulfuric acid reagent.405 The protonation of the bipyrimidine ligand weakens the Pt-N bonds, but the ligand remains coordinated to Pt.407 In addition, the L/OSO3H- ligand exchange is exoergic for L ) NH3, with ∆G ) -23 kcal mol-1, but endoergic for L ) bipym, with ∆G ) 16 kcal mol-1. All these results indicate that NH3 is easily dissociated from [Pt(NH3)2(Cl)2] in hot concentrated H2SO4, leading to PtCl2, which will induce catalyst deactivation by the formation of a [PtCl2]n precipitate. The influence of the pH on the oxidation of methane by [Pt(bpym)(Cl)2]/H2SO4 was further explored in a subsequent study.409 Two different prototropic forms of the Pt(II) active species, [Pt(Hbpym)(Cl)(OSO3H)]+ (Hbpym ) protonated bipyrimidine) and [Pt(Hbpym)(Cl)(OSO3H2)]2+, were studied. In Hbpym, a nitrogen not bound to Pt is protonated, so chelation is preserved. Both OSO3H- and OSO3H2 act as η1 ligands bound to Pt through O. The dicationic complex, which is generated by the protonation of the OSO3H- ligand, will predominate under the strong acidic conditions of the reaction. The most difficult step in the CsH activation process is methane uptake, in which CH4 coordinates to Pt by displacing the S-containing ligand. The calculations showed that the energy barrier associated with this step is significantly reduced by protonation, from 39.0 kcal mol-1 for [Pt(Hbpym)(Cl)(OSO3H)]+ to 27.9 kcal mol-1 for [Pt(Hbpym)(Cl)(OSO3H2)]2+. These results suggest that sulfuric acid acts not only as an oxidant but also as a cocatalyst. Periana and Goddard improved the efficiency of the Catalytica sytem by exploiting the interplay between experiments and theory.410 The calculations showed that the CsH activation of methane involves an overall energy barrier of ∼32 kcal mol-1, which is too high to make the process commercially viable. This energy barrier can be separated into 27 kcal mol-1 for methane coordination and 5 kcal mol-1

Balcells et al.

Figure 89. Cyclic six-membered transition state for the CsH activation of benzene catalyzed by [Pt(pic)(TFA)2]-.

for CsH cleavage. Hence, the efficiency of the catalyst would be improved by promoting methane coordination without proportionally increasing the barrier associated with CsH cleavage. The calculations predicted that these requirements would be fulfilled by the [Pt(pic)(TFA)2]- (pic- ) η2-N,O-picolinate, TFA- ) trifluoroacetate) catalyst. In contrast with the neutral bpym ligand, the monoanionic character of the pic- ligand should facilitate methane coordination by making the metal center more electron-rich. In addition, the TFA- ligand has the ability to assist the CsH cleavage step. These hypotheses were confirmed by comparing the catalytic properties of [Pt(bpym)(TFA)2] and [Pt(pic)(TFA)2]- in the oxidation of benzene with H2SO4. The experiments showed that the [Pt(pic)(TFA)2]- catalyst is 300 times more active than [Pt(bpym)(TFA)2]. With the bpym ligand, the coordination of benzene involves an energy barrier of 14 kcal mol-1, which is reduced to 5 kcal mol-1 with the pic- ligand. In contrast, the barrier for CsH cleavage is increased by 3 kcal mol-1 when the pic- ligand is used. Nevertheless, the overall CsH energy barrier is reduced from 27 kcal mol-1 for [Pt(bpym)(TFA)2] to 21 kcal mol-1 for [Pt(pic)(TFA)2]-, which is not only lower but also in good agreement with the experimental value, 23 kcal mol-1. The CsH cleavage is assisted by the TFA- ligand, which takes a proton from the activated bond yielding a cyclic sixmembered transition state (Figure 89). One of the main drawbacks of the Catalytica process is deactivation by water. However, Cheng et al. reported that tolerance to water is increased by adding pyrazolium-based ionic liquids.411 A theoretical study by Oxgaard and Goddard showed that in the presence of these additives, the solid PtCl2 catalyst is dissolved in concentrated H2SO4 due to the formation of a strong ion pair, [PtCl4][HPzm]2 (HPzm ) protonated pyrazolium).412 The formation of this ion pair, in which the [PtCl4]2- and HPzm+ ions are bound by electrostatic forces and H · · · Cl H-bonds, is exothermic by 68.3 kcal mol-1. Interestingly, the substitution of a Cl- ligand of the ion pair by either HSO4- or H2O is energy-uphill, in contrast with the original Catalytica catalyst, [Pt(bpym)(Cl)2], in which it is exothermic. This indicates that the ionic liquid prevents the poisoning by water. The CsH activation of methane by [PtCl4][HPzm]2 follows a two-step mechanism: (1) coordination of CH4 to Pt by an associative reaction, in which a Cl- ligand is displaced toward the second coordination sphere with the assistance of the HPzm counterions, and (2) CsH bond cleavage by oxidative addition yielding a cisPt(CH3)(H) intermediate. The transition states associated with these reactions are 26.5 kcal mol-1 for methane-coordination and 21.1 kcal mol-1 for CsH cleavage above the energy of reactants. This study showed that the introduction of ionic liquids in the Catalytica process not only reduces the deactivation of the catalyst by water, but also facilitates methane activation and changes the nature of the CsH cleavage mechanism from ambiphilic metal-ligand activation (AMLA) to oxidative addition.

CsH Bond Activation in Transition Metal Species

The tolerance of the Shilov reaction to water has been also enhanced by Vedernikov, who reported the synthesis and reactivity of [Pt(X)(CH3)2]-.413,414 These complexes, in which the Pt(II) center is chelated by a bidentate anionic dipyridine ligand (X), promote the CsH activation of benzene in biphasic water-benzene systems.414 DFT calculations suggest that the protonation of [Pt(X)(CH3)2]- in the water phase generates a lipophilic [Pt(X)(CH3)2H] species, which is efficiently extracted into the benzene phase. This complex generates [Pt(X)(CH3)(CH4)] in a facile reductive elimination step, with ∆Hq ) 9.2 kcal mol-1 and ∆H ) 5.2 kcal mol-1. Since the organic phase is ∼100% benzene, the methane ligand is easily replaced by benzene, which yields [Pt(X)(CH3)(C6H6)]. In a subsequent step, one of the CsH bonds of the C6H6 ligand is cleaved by oxidative addition, leading to a Pt(IV) hydride intermediate, [Pt(X)(CH3)(C6H5)(H)], which completes the CsH activation process. This study shows that [Pt(X)(CH3)2H] is an active species in arene CsH activation. The abstraction of this complex into the reactant phase allows for high efficiency even in the presence of water, which remains in the other phase of the system. A new computational approach to the discovery of new catalysts, known as quantum mechanical rapid prototyping (QM-RP), was developed by Goddard.415 The QM-RP approach profits from the excellent accuracy/cost relationship offered by modern QM calculations, thanks to the DFT methods, and follows a five-step flow diagram: (1) complete and detailed description of the reaction mechanism for a wellknown experimental system, in which all intermediates and transition states involved in the mechanism are computed; (2) characterization of the catalytic bottlenecks, in which the critical steps involving the largest activation or reaction energies are identified; (3) discovery of new catalysts by screening, in which a variety of modifications in the nature of metal, ligand, cocatalyst, or solvent are applied in order to test how they affect the catalytic bottlenecks and other critical factors like catalyst stability and poisoning; (4) catalyst refinement, in which the leads found in the previous step are further improved by fine-tuning their electronic and steric properties; (5) experimental tests, in which the leads are finally synthetized and tested in the laboratory. The QM-RP approach was applied to the oxidation of methane by the Catalytica system in the quest for new and more efficient catalysts.415 The calculations on the nature of the reaction mechanism revealed the strong influence of the pH on the reactivity of the [Pt(bpym)(Cl)2] catalyst. The cleavage of the CsH bond is thermodynamically and kinetically favored by the protonation of the bpym ligand. In addition, it was found that water inhibits the reaction by solvating the ground state of the reactants, which is more stabilized than the subsequent intermediates and transition states. By application of the QM-RP protocol, other metals, including Ir, Os, and Au, and ligands, including tridentate NNN, NNC, and NON and bidentate NN and OO ligands, were tested. One of the new catalysts found in this study was an Ir(III) complex with a pincer NNC ligand, which was predicted to promote CsH activation through energy barriers lower than 30 kcal mol-1, without being poisoned by water. Six years later, Periana and Goddard reported the synthesis of this complex, which turned out to be an excellent catalyst for methane oxidation in trifluoroacetic acid.416 A variation of the original Catalytica process, based on a new gold catalyst, was developed by Periana.417 In the

Chemical Reviews, 2010, Vol. 110, No. 2 799

presence of Au(0), methane is oxidized to methanol with selectivities higher than 90%. The reaction is performed in a 3 M solution of H2SeO4, which is needed to oxidize and solubilize Au(0), in 96% H2SO4 at 180 °C. Under these conditions, the catalyst is formed in situ and can be either a Au(III) or a Au(I) species. DFT calculations considering solvent effects showed that the most stable complexes are the square planar [Au(SO4)2]- for Au(III) and the linear [Au(OSO3H)2]- for Au(I). In the case of [Au(SO4)2]-, one of the SO42- ligands needs to be protonated to facilitate the coordination of methane, which undergoes ambiphilic metal-ligand CsH activation. This step has an energy barrier of 28.1 kcal mol-1 and involves a proton transfer from methane to the OSO3H- ligand. The final functionalization step, which has a much lower energy barrier, consists of an SN2-like reaction in which a free bisulfate anion attacks the CH3 ligand. The alternative Au(I) catalyst, [Au(OSO3H)2]-, promotes CsH cleavage by oxidative addition with an energy barrier of 21.1 kcal mol-1, lower than that associated with the Au(III) catalyst. These results suggest that [Au(OSO3H)2]- is catalytically more efficient than [Au(SO4)2]-. Nevertheless, the small energy difference, 7 kcal mol-1, between the highest energy barriers found in each case and the concentration of Au(I), which is unlikely to be much higher than that of Au(III), suggest that both complexes may act as active species. Periana also reported that Pd2+ dissolved in concentrated H2SO4 catalyzes the direct oxidation of methane to acetic acid in the presence of O2 and CO at 453 K.418 This reaction was computationally studied by Bell.419 Interestingly, the model used in this study was built by assuming that Pd(II) is totally solvated by H2SO4, so all the ligands in the first coordination sphere are either OSO3H- or OSO3H2. The calculations indicated that the dominant species in the Pd(II)/ H2SO4 mixture is [Pd(η2-OSO3H)(OSO3H)(OSO3H2)]. In this square planar Pd(II) complex, one of the hydrogen sulfate ligands, η2-OSO3H-, is bound to Pd through two of its oxygen atoms, whereas in the other ligands only one oxygen is bound to the metal. This species undergoes a concerted metalation deprotonation (CMD) of methane, in which the formation of the PdsCH3 bond is concomitant with a proton transfer to the η2-OSO3H- ligand. This step has the highest activation barrier, ∆Gq ) 41.5 kcal mol-1, and leads to the formation of [Pd(CH3)(OSO3H)(OSO3H2)2]. In the following step, CO adds and inserts into the PdsCH3 bond, involving a much lower activation barrier of 14.7 kcal mol-1. The final reductive elimination of CH3COOSO3H, which yields the final acetic acid product by hydrolysis, requires the previous oxidation of Pd(II) to Pd(IV) by H2SO4 and O2. The oxidation and reductive elimination steps are exothermic and involve very low energy barriers. These results indicated that Pd(II) is capable of promoting methane oxidation in the absence of chloride, amine, and aqua ligands. The calculations also suggested that Pd(II) is reduced to Pd(0) by CO, which causes the death of the catalyst due the formation of a Pdblack precipitate. Another alternative to the Shilov and Periana catalysts is [Rh(CO)2(I)2]-. Sen found that this complex, which is also used in the Monsanto process, catalyzes the oxidation of methane to methanol and acetic acid by oxygen and carbon monooxide under high pressure.420,421 The mechanism of this reaction was determined in a theoretical study by Hristov and Ziegler (Figure 90).422 In the first step, one I- ligand is substituted by CH4 through a two-step dissociative ligand

800 Chemical Reviews, 2010, Vol. 110, No. 2

Figure 90. Key steps in the oxidation of methane by [Rh(CO)2(I)2]-.

Figure 91. Methane oxidation by permanganate.

exchange mechanism, with an energy cost of only 7.8 kcal mol-1. In the subsequent CsH cleavage step, the oxidative q ) 24.8 kcal mol-1, is favored addition mechanism, ∆Gsol q over σ-bond metathesis, ∆Gsol ) 29.8 kcal mol-1. The complex obtained in this step, [Rh(CO)2(I)(CH3)(H)], eliminates the hydride ligand as a proton by reacting with I-, which leads to the formation of [Rh(CO)2(I)(CH3)]-. This species can undergo either hydrolysis, which yields the methanol product, or CO addition and subsequent insertion into the RhsCH3 bond. The latter process, which yields the acetic acid product after hydrolysis, is the most favorable one, involving an energy barrier of 14.7 kcal mol-1. This is in good agreement with the experiments, which show that acetic acid is the major product. The calculations also suggest that the selectivity is switched to methanol in more acidic conditions because the solubility of CO is reduced at low pH.

5.2. Metal-Oxo Species 5.2.1. Inorganic Reagents The reaction mechanism for the oxidation of methane by permanganate was explored by Strassner and Houk using a DFT method423 (Figure 91). CsH activation starts with a H abstraction step leading to the formation of a radical ion pair intermediate, Mn(OH)O3•-/CH3•. This species collapses into the final reaction product, Mn(OH)(OCH3)O2-, with no energy barrier. In the transition state, two oxygen atoms of permanganate interact with methane, one with the H being transferred and the other one with the carbon. The final step differs from the classical rebound mechanism, because the methyl radical is trapped by a MndO group yielding MnsOCH3, rather than forming CH3OH by OH rebound. The reaction has a significant energy barrier and is clearly endothermic, with ∆Hq ) 32.3 kcal mol-1 and ∆H ) 24.3 kcal mol-1. The same mechanism was characterized for the oxidation of the benzyl carbon of toluene by MnO4-. In this

Balcells et al.

Figure 92. Coordination (1) and rebound (2) mechanisms in methane-to-methanol oxidation by the Fenton reagent.

case, the energy barrier was lower than that of methane by 10.5 kcal mol-1. It was also shown that radical character is present in this reaction. Alkanes and arylalkanes can also be oxidized using KMnO4 in acetonitrile. The rate of this process becomes faster by over 7 orders of magnitude if a few equivalents of BF3 are added to the reaction mixture. Experimental and computational data424 indicate that BF3 reacts with MnO4yielding an adduct, [(BF3)xMnO4]- (x ) 1, 2), which is the active species in CsH oxidation. The kinetic isotope effects, kH/kD ≈ 7 for toluene and ethylbenzene, and the correlation between log(k) and the bond strength, D(CsH), suggest that the rate-determining step of the reaction is a H atom transfer from the substrate to the MndO moiety. DFT calculations show that BF3 accelerates the reaction by lowering the energy barrier associated with this process, from 32.1 kcal mol-1 for MnO4- to 26.2 kcal mol-1 for [BF3MnO4]- and to 20.5 kcal mol-1 for [(BF3)2MnO4]-. The analysis of the electronic structure showed that BF3 lowers the energy barrier by helping the metal to reach a Mn(VI) d1 configuration in the transition state. The oxidation of methane to methanol by the Fenton reagent was theoretically studied by Baerends.425 In previous studies,426,427 it was demonstrated that the solvated ferryl ion, [Fe(O)(H2O)5]2+, is easily formed from the Fenton reagent, which is an aqueous mixture of Fe2+ and H2O2 used as oxidant. Two different mechanisms were explored in Vacuo by means of static DFT calculations: the oxygen rebound mechanism postulated for cytochrome P450 and the coordination mechanism postulated for bare metal oxo ions, which requires a free coordination site at the metal center (Figure 92). In both mechanisms, the cleavage of the CsH bond is associated with the highest energy barrier: 3.4 kcal mol-1 for the rebound mechanism and 22.8 kcal mol-1 for the coordination mechanism. Significant preference was thus found for the rebound mechanism. In the presence of the solvent, water, which was modeled by means of CPMD calculations, the energy barrier for H abstraction in the rebound mechanism is increased to 21.8 kcal mol-1. The barrier in solution is much higher than that in vacuo due to the decreased hydration of the reacting oxo/hydroxo ligand and the higher energy of solvation of the FeIVO ion with respect to the FeIIIOH ion. This study strongly supports the [Fe(O)(H2O)5]2+ complex as the active species in the oxidation of organic compounds by the Fenton’s reagent.

CsH Bond Activation in Transition Metal Species

Chemical Reviews, 2010, Vol. 110, No. 2 801

Figure 94. The oxidation of adamantane can yield either 1- or 2-adamantol.

Figure 93. Direct (1) and coordination (2) mechanisms in methanol-to-formaldehyde oxidation by the Fenton reagent.

The product of the first oxidation reaction, methanol, can be further oxidized by the Fenton reagent to formaldehyde. The mechanism of this reaction was also studied by Baerends at the DFT level.428 The reaction was modeled both in the gas phase, by considering microsolvated clusters, and in aqueous solution by means of CPMD calculations. Two different mechanisms with variants were investigated (Figure 93): a direct mechanism, in which methanol is directly attacked by the FeIVO2+ moiety, and a coordination mechanism, in which the OH group of methanol coordinates previously to the metal. The reaction does not follow the oxygen rebound mechanism; after the first step, the radical intermediate undergoes a second H abstraction yielding an FeII(OH2)2+ species, rather than OH rebound. In the direct mechanism, the first H abstraction involves the CsH bond, whereas in the coordination mechanism, it involves the OsH bond. The energy barrier associated with the homolytic CsH cleavage in the direct mechanism is dramatically increased from less than ca. 1 kcal mol-1 in gas phase to ca. 12 kcal mol-1 in solution, as in the case of methane-to-methanol oxidation.425 The subsequent H abstraction from the OsH bond proceeds spontaneously without energy barrier. In the coordination mechanism, the first H abstraction involves an energy barrier of more than 25 kcal mol-1 in the gas phase. This barrier is lowered to 8-12 kcal mol-1 by the participation of bridging molecules of water, which assist the transfer of the H from the O of methanol to the O of the FeIVO2+ moiety. Both mechanisms are thus possible. Nevertheless, the coordination mechanism requires a previous ligand substitution reaction, in which one aqua ligand is replaced by methanol. The calculations also showed that the deprotonation of [Fe(O)(H2O)5]2+ to form [Fe(O)(H2O)4(OH)]+ decreases the oxidizing power of the Fenton reagent by increasing the associated energy barriers. These results suggest that optimal reactivity is achieved under acidic conditions, in total agreement with experiments. CsH oxidation is promoted by several metal-oxo moieties, but [FeIVO]2+ is by far the most widespread. [FeIVO]2+ is the reactive core in the Fenton reagent, in biomimetic iron complexes, in heme and non-heme enzymes, and in iron-

containing zeolites. In general, metal-oxo complexes act as potent electrophiles capable of accepting electrons from very poor donors, like aliphatic CsH bonds. Theoretical studies by Baerends429–431 showed that in the case of [FeIVO]2+, both the energy and the shape of the lowest acceptor orbital depend on the spin state and the nature of the ligands and the solvent. The role of the metal was also clarified by Baerends with a systematic study on the oxidation of methane to methanol in the gas phase by [MO(H2O)x]2+ (M ) V, Cr, Mn, Fe, or Co, x ) 5, and M ) Ni, Cu, x ) 4).432 For all metals, the reaction follows the oxygen rebound mechanism, with the H abstraction step having the highest energy barrier. The calculations show that the energy of the lowest acceptor orbital is crucial; the barrier associated with the H abstraction step is lowered as this energy decreases. In addition, the nature of the orbital plays also a key role; at a given energy level, a σ* acceptor is much more efficient than a π* acceptor. [FeIVO]2+ is thus specially efficient in promoting CsH oxidation because it has a low-lying σ* acceptor ligand. Nevertheless, the calculations suggest that the abundance of iron is likely to be a relevant factor, because [FeIVO]2+ is particularly efficient but [MnIVO]2+ and [CoIVO]2+ are also potent oxidants. The competitive oxidation of adamantane to 1- and 2-adamantol by ferrate(VI) was studied at the DFT level by Yoshizawa,433 considering the different protonation states of the oxidant, FeO42-, HFeO4-, and H2FeO4 (Figure 94). Diprotonated ferrate, H2FeO4, is the best oxidant, promoting hydroxylation by the classical rebound mechanism. The energy barrier of H abstraction was found to be lower for the formation of 1-adamantol (∆Eq ) 6.9 kcal mol-1) than for the formation of 2-adamantol (∆Eq ) 8.4 kcal mol-1), showing a preference for the cleavage of a tertiary over a secondary CsH bond. Transition state theory was applied successfully to assess the temperature dependence of the CsH bond oxidation rates and the ratio of 1- versus 2-adamantol products. The selectivity of CsH oxidation by the rebound mechanism is controlled by the relative stability of the involved radicals; the reaction pathway involving a tertiary radical (1-adamantol) is favored over that involving a secondary radical (2-adamantol).

5.2.2. Non-heme Iron Biomimetics Several non-heme enzymes, including methane monooxygenase and ribonucleotide reductase, catalyze the oxidation of CsH bonds. The active centers of these enzymes, which contain the oxoferryl group, FeIVdO, have been modeled by means of the synthesis of biomimetic complexes.226,227 These complexes also play a crucial role in the quest for efficient synthetic catalysts active in alkane oxidation.434 In the iron non-heme complexes, the [FeIVO]2+ cation is stabilized by a polydentate chelating ligand in an octahedral coordination environment. Some of the most relevant ligands used in the synthesis of these complexes are represented in

802 Chemical Reviews, 2010, Vol. 110, No. 2

Figure 95. Polydentate chelating ligands used in the synthesis of iron non-heme biomimetic complexes.

Figure 95. When the ligand is tetradentate, TPA and TMC, the remaining vacant site is occupied by a molecule of solvent, that is, MeCN, or a counteranion, in most cases coordinated trans to the oxo group. These species have been studied by a combination of spectroscopic and theoretical methods. The calculations have shown that reactivity depends strongly on the nature of the spin states of the active FeIVdO complexes. Alkene hydroxylation catalyzed by [Fe(TPA)] (TPA ) tris(2-pyridyl-methyl)amine) was studied by Bassan with a DFT method.435 In the presence of H2O2, this biomimetic non-heme complex generates [Fe(O)(TPA)(OH)], which is believed to be the active species. In [Fe(O)(TPA)(OH)], an oxo and a hydroxo ligand are coordinated in cis to the Fe(V) center in an octahedral environment. The quartet state is more stable than the doublet, but the calculations suggested that quartet to doublet spin crossover may be involved in the reaction. The calculations showed that methane and propane are hydroxylated by [Fe(O)(TPA)(OH)] following the classical rebound mechanism. In both cases, the barrier for radical H abstraction is clearly higher than that of OH rebound. The barrier in the case of methane, 17.0 kcal mol-1, which involves the cleavage of a primary CsH bond, is higher than that for propane, 7.0 kcal mol-1, involving a secondary CsH. The rebound barrier is extremely low, and in the case of propane, the relaxation of the H abstraction transition state led directly to the final 2-propanol product. This suggests that the radical intermediate has a very short lifetime, which agrees with the stereospecificity of the reaction observed experimentally. An oxo-hydroxo tautomerism mechanism was also computed. The tautomerism involves a water-assisted concerted proton transfer from the oxo group to the hydroxo, with an energy barrier of 14.2 kcal mol-1. This mechanism accounts for the incorporation of oxygen coming from the solvent, water, into the final reaction product observed experimentally. The oxidation of acetonitrile, which is a common solvent in this reaction, was also explored. This process is exothermic but involves a

Balcells et al.

relatively high energy barrier, 18.3 kcal mol-1, which makes this process unlikely, in good agreement with experiments. In a DFT study on methane oxidation by [Fe(O)(TPA)(OH)]2+, Balbuena demonstrated that the catalytic activity of this compound can be tuned by introducing different substituents in the TPA ligand.436 Within the framework of the rebound mechanism, the calculations show that the energy barrier to H abstraction for TPA, 15.7 kcal mol-1, is lowered to 13.4 kcal mol-1 by introducing nitro groups in the pyridine rings and is increased to 18.0 kcal mol-1 by introducing methoxy groups in the same position. These results indicate that the critical barrier of the reaction is lowered by electron-withdrawing groups whereas it is increased by electron-donating groups. This is consistent with a reactivity model in which the catalyst, acting as the electrophile, undergoes nucleophilic attack of the CsH bond. CsH hydroxylation by the biomimetic non-heme [Fe(O)(N4Py)]2+ (N4Py ) N,N-bis(2-pyridylmethyl)-N-bis(2pyridyl)methylamine) complex was studied by Shaik with a DFT method including solvent effects.437 This compound acts as a potent oxidant capable of activating inert CsH bonds, with dissociation energies higher than 95 kcal mol-1. The ground state of [Fe(O)(N4Py)]2+ is a triplet, in agreement with experiments, followed by a quintet state and an openshell singlet. The rebound mechanism for the hydroxylation of cyclohexane, which has a strong CsH bond (D(CsH) ) 99 kcal mol-1), was explored in detail. The calculations showed that the H abstraction barrier is 9.6 kcal mol-1 in vacuo, which is increased to 11.9 kcal mol-1 by solvation and to 26-27 kcal mol-1 by entropic effects. These results suggested that [Fe(O)(N4Py)]2+ is intrinsically a more potent oxidant than the compound I species of CYP450 due to the double positive charge of the system, which stabilizes the three-center three-electron (3c/3e) O · · · H · · · C moiety of the transition state. These studies were subsequently extended by exploring the mechanism of alkane hydroxylation by several iron oxo complexes. Three substrates, toluene, cyclohexane, and acetonitrile, and three FeIVO oxidants with different N-bound chelating ligands, TMC (1,4,8,11-tetramethyl-1,4,8,11-tetraazacyclotetradecane), N4Py, and Bn-TPEN (N-benzylN,N′,N′-tris(2-pyridylmethyl)-1,2-diaminoethane), were considered.438 The calculations provide a rationalization of the experimental reactivity trends by means of a two-state reactivity (TSR) model. In this model, two spin states, which must be close in energy, interplay giving rise to the overall reactivity observed experimentally. The FeIVO biomimetics have a triplet ground state and a low-lying quintet state, within an energy range of 1-10 kcal mol-1. Both spin states are able to promote CsH oxidation but involve different reaction mechanisms. In the triplet state, the reaction follows the classical stepwise oxygen rebound mechanism, in which H abstraction is followed by the rebound of the carbon radical to the OH group. In the quintet state, the reaction is effectively concerted because the OH rebound step is barrier free. Another key feature is the fact that the quintet is less stable but more reactive than the triplet, because in the quintet state the energy barrier associated with the rate-determining H abstraction step is lower. Depending on the energy difference between the spin states, ∆ET/Q, which is determined by the nature of the ligands bound to Fe, two main TSR scenarios are possible (Figure 96): (1) when ∆ET/Q is small, the two states undergo pre-equilibration and the subsequent reaction takes place in the quintet state, which

CsH Bond Activation in Transition Metal Species

Figure 96. The TSR model can involve either triplet/quintet preequilibration (case 1) or triplet to quintet spin crossover en route to the transition state (case 2).

is associated with the lowest energy path, or (2) when ∆ET/Q is large, spin-crossover from triplet to quintet may happen en route to the H abstraction transition state. Between these extremes, several intermediate cases are possible. In addition, in case 2, if the triplet and quintet transition states are close in energy or the spin-crossover is ineffective, the reaction may take place entirely on the triplet surface. This will have a dramatic effect on the selectivity of the reaction; in the triplet state, the generation of radicals will lower the selectivity due to the formation of side products or the loss of stereochemistry. The TSR model, previously proposed by Shaik for cytochrome P450 on the basis of calculations,439–441 has allowed rationalization of several intriguing aspects of the chemistry of the non-heme iron biomimetics.442–445 Surprisingly, experiments show that the oxidation of phosphines and alkanes by the series of complexes [TMC(L)FeO]2+ (L ) CH3CN, CF3CO2-, N3-, SR-) follows opposite trends. The rate of phosphine oxidation increases as the electrophilicity of the oxidant becomes higher, following the order CH3CN > CF3CO2- > N3- > SR-, whereas alkane hydroxylation follows the counterintuitive opposite trend. The TSR model provides a simple explanation for this unexpected observation.442,443 As the electron donor character of L decreases, the FeIVO complex becomes more electrophilic and thus more oxidating, but the quintet state is shifted up in energy. Therefore, the participation of the triplet state, which involves higher energy barriers, increases, and the overall reactivity is diminished. This counterintuitive trend is not observed for the RuIVO analogues, which show lower reactivities. In the framework of the TSR model, these experimental results are explained by the high energy of the quintet state, obtained when Fe is replaced by Ru.444 Due to this high energy, the quintet is not accessible, and the system follows a single state reactivity model, which involves only the triplet state. Without the interplay between the different spin states, the relationship between the electrophilicity and the reactivity of the RuIVO complexes follows the expected trend for both phosphine and CsH oxidation. In addition, the high energy barriers associated with the triplet state account for the sluggish reactivity of these Ru species. Another intriguing feature is the unusual kinetic isotope effect pattern measured for [Fe(O)(N4Py)]2+ and [Fe(O)(Bn-TPEN)]2+. For some substrates, like acetonitrile and cyclohexane, the KIE falls within the semiclassical region, with kH/kD < 7. Conversely, other substrates, like ethylbenzene and dihydroanthracene, show large nonclassical KIE, with kH/kD > 7, in some cases g30. In the latter case, the large KIEs are typical of tunneling and support the homolytic

Chemical Reviews, 2010, Vol. 110, No. 2 803

CsH cleavage as the rate-determining step. Nevertheless, if only the triplet participates in the reaction, the kH/kD values are too large, considering the high energy barriers associated with this spin state. In contrast, in the TSR framework, the system may cross the high triplet barrier by hoping to the lower energy quintet surface, opening a tunneling-like pathway.445 If the zero-point energy stored in the CsH bond that undergoes activation is equal to or higher than the quintet energy barrier, the vibration of this bond may propagate the reactant directly to the products. This phenomenon depends strongly on the strength of the CsH bond, D(CsH). As D(CsH) increases, the quintet energy barrier becomes higher thus hampering the tunneling-like pathway. Therefore, the large KIEs are only possible for a narrow range of substrates with weak CsH bonds, D(CsH) < 93 kcal mol-1, as observed experimentally for ethylbenzene and dihydroanthracene. With stronger CsH bonds, the KIEs follow the semiclassical regime. The effect of external electric fields (EF) on CsH oxidation by non-heme iron-oxo complexes was studied by Shaik at the DFT level.446 The application of an EF changes the potential energy surface of the system, which has a great impact on the mechanism, the selectivity, and the rate of the reaction. The largest effects are observed when the EF is aligned with the axis of the FedO bond, in which the electrons flow from the substrate to the Fe center during the reaction. This electron flow is stimulated by the EF yielding lower energy barriers. Two different spin states were studied, the triplet and the quintet, and the effect of the EF was found to be larger in the quintet state. The influence of the EF on the reaction mechanism was explored considering two different ligands bound to Fe, N4Py and TMC(SR), and two different substrates, cyclohexane and toluene. When the iron complex is a good acceptor, [Fe(O)(N4Py)]2+, the reaction did not follow the classical OH rebound pathway. If the substrate is a good acceptor, toluene, the H abstraction step is replaced by electron transfer, whereas if it is a poor acceptor, cyclohexane, the EF promotes hydride transfer yielding a carbocation. When both the acceptor and donor characters are poor, like in the reaction of [Fe(O)(TMC(SR))]+ with cyclohexane, the rebound mechanism prevails and the EF lowers the energy barriers of the H abstraction and the OH rebound steps. The [Fe(O)(TMCS)]+ (TMCS ) 1-mercaptoethyl-4,8,11trimethyl-1,4,8,11-tetraazacyclotetradecane) complex is an interesting biomimetic of the cytochrome P450 enzyme. This complex promotes the CsH oxidation of several alkanes in a chemoselective fashion. For instance, when [Fe(O)(TMCS)]+ is used to oxidize propene, which can easily undergo epoxidation of the CdC bond, only the CH3hydroxylation product is obtained. This selective reaction was theoretically studied by de Visser.447 The calculations showed that the reaction follows a single-state reactivity regime, since both the hydroxylation and the epoxidation take place fully in the quintet state. The hydroxylation process follows the rebound mechanism with the highest energy barrier, 15.0 kcal mol-1, being associated with the H abstraction step. This barrier is 3 kcal mol-1 lower than the highest energy barrier associated with the epoxidation reaction, suggesting that hydroxylation takes place preferentially, in total agreement with the experiments. The selectivity is due to the steric hindrance between the catalyst and the substrate in the transition state, which is larger in the epoxidation reaction. The comparison of [Fe(O)(T-

804 Chemical Reviews, 2010, Vol. 110, No. 2

Figure 97. The π-FMO (left) and the σ-FMO (right) pathways in CsH oxidation by non-heme FeIVdO biomimetic complexes.

MCS)]+ with the natural enzymatic systems revealed that the biomimetic complex involves higher energy barriers, but is less sensitive to external perturbations like the H bonds or the dielectric constant associated with the solvent. The series of non-heme FeIVdO complexes [Fe(O)(TMC)(NCMe)]2+, [Fe(O)(TMC)(OCOCF3)]+ and [Fe(O)(N4Py)]2+ were studied by means of spectroscopic and DFT techniques by Que and Solomon.448 The ground state of these complexes is a low-spin triplet, in contrast with the Fe enzyme intermediates, which are high-spin quintet species. The FedO bonds are strong and have high covalent character. The complex with the strongest FedO bond, [Fe(O)(N4Py)]2+, has also the highest reactivity in CsH oxidation. Due to the strength and covalency of the iron-oxo bond, the frontier molecular orbitals (FMOs) that govern the reactivity have strong oxygen character. These FMOs, which are the antibonding π*xz/yz(FedO) orbitals, dxz-px and dyz-py, activate the iron complex for an electrophilic attack. In the transition state of the key H abstraction step, the σ(CsH) orbital of the oxidized CsH bond interacts with one of the π*xz/yz(FedO) orbitals (Figure 97). In this reaction pathway, labeled as π-FMO, there is a lateral approach of the substrate perpendicular to the FedO direction, with an Fe-O-H angle of ∼120° for an optimal overlap between the σ(CsH) and π*xz/yz(FedO) orbitals. This π-FMO pathway is also followed by the Fe enzyme intermediates. In these systems, the σ*z(FedO) orbital, dz2-pz, is also low in energy and has high oxygen character, due to the spin polarization associated with the high-spin quintet ground state. This orbital opens an alternative σ-FMO pathway in which the substrate approaches the iron complex in a vertical orientation along the FedO direction. Two different FeIVdO complexes, one heme ([Fe(O)(por)(NCMe)]2+ (por ) porphyrin)) and one non-heme ([Fe(O)(TMC)(NCMe)]2+), were studied and compared by Solomon.449 The electronic structure and the iron-oxo bonding are essentially the same in both complexes, as indicated by the very similar FMOs and Fe-O distances and vibrational frequencies. Thus, both the porphyrin and the TMC ligands do not interact to a great extent with the π FedO bond. This is normal for the TMC ligand, which is a pure σ-donor, but it is not expected for the π-donor/acceptor porphyrin ligand. The calculations show that this is due to the strong and covalent character of the iron-oxo bond, which decouples the π system of the FedO group from the π system of the porphyrin. Nevertheless, in CsH oxidation reactions, the product of the key H abstraction step, which is an FeIIIOH intermediate, has different electronic structure and Fe-O bonding in the heme and non-heme environments. The weakening of the iron-oxygen bond along the reaction pathway from FedO to Fe-OH promotes the mixing of the π systems of the reactive center and the ligand in the product, which is only possible with the porphyrin ligand of the heme complex. As a consequence, the FeIIIOH intermediate is more stable with the porphyrin ligand than with TMC, and

Balcells et al.

therefore, H abstraction is thermodynamically more favorable in the heme case. The non-heme iron(II) [Fe(TpPh2)BF] (TpPh2 ) hydrotris(3,5-diphenylpyrazol-1-yl)borate; BF ) benzoylformate) complex promotes the hydroxylation of phenyl rings, using O2 as oxidant. This complex reproduces the activity of the R-ketoglutarate-dependent enzymes. A DFT study reported by Borowski and Siegbahn showed that this reaction does not follow the classical oxygen rebound mechanism.450 The oxidation of [Fe(TpPh2)BF] by O2 yields a high-valent FeIVO species (Figure 98). The oxo group makes an electrophilic attack to a CsH bond of the phenyl ring, generating a new C(H)sO bond. In the next step, the hydrogen is transferred to O with the assistance of the BF ligand, yielding the CsOH group of the phenol product. A single electron is transferred in each step; the formation of the CsO bond leads to a phenyl radical, and the H transfer is indeed a proton-coupled electron transfer. As in the multistate reactivity models proposed by Shaik, three different spin states may participate in the reaction, a triplet, a quintet, and a septet. This reaction mechanism is also followed in aromatic hydroxylation of arenes by the FeIVO2+ complexes [Fe(O)(Bn-tpen)]2+ and [Fe(O)(N4Py)]2+. This reaction was analyzed in a combined experimental/theoretical study by Nam and de Visser.451 Isotope labeling experiments proved that the oxygen of the product comes from the [FeIVO]2+ moiety, thus confirming the iron-oxo complexes as active species. The large and negative value of the Hammet F parameter, -3.9, and the inverse KIE value of 0.9, were consistent with an electrophilic pathway. This was further confirmed by the calculations, which indicated that the reaction follows an electrophilic substitution mechanism. The reaction starts with an electrophilic attack of the oxidant to a carbon of the aromatic ring, leading to an FesOsPh intermediate, in which the phenyl ring can have either radical or cationic character. This step is followed by a proton-shuttle mechanism in which the hydrogen at the ipso position is transferred to the oxygen with the help of a N of the ligand, yielding the final phenol product. This mechanism is more efficient than H abstraction followed by OH rebound. The calculations show that the TSR model also applies to this reaction, which involves both the triplet and quintet states. The non-heme biomimetic complex [Fe(O)(R-TPEN)]2+ has three different isomers, labeled A, B, and C. In isomers A and B, the three pyridine rings of the R-TPEN ligand are cis to the oxo group, whereas in isomer C one pyridine is in trans position. Isomers A and B differ in the orientation of the pyridine rings with respect to the FedO axis; one pyridine is perpendicular to this axis in A and two in B. The hydroxylation of cyclohexane by the three isomers of [Fe(O)(R-TPEN)]2+ was studied by Han452 The calculations showed that isomers A, B, and C have triplet ground states and low-lying quintet states. The three isomers promote CsH oxidation. The reaction follows the oxygen rebound mechanism and involves both the triplet and quintet states, within a TSR framework. The relative stability of the isomers follows the order A > B > C, in good agreement with previous studies by Klinker et al.453 Conversely, the reactivity follows a different trend, B > A > C. The higher reactivity of isomer B with respect to A, is due to the presence of one more pyridine ring perpendicular to the FedO axis, which stimulates the reactivity of the iron-oxo

CsH Bond Activation in Transition Metal Species

Chemical Reviews, 2010, Vol. 110, No. 2 805

Figure 98. The reaction mechanism of phenyl hydroxylation by FeIVO.

Figure 99. Halogenase biomimetics: the two possible non-heme iron-oxo active species.

group. The presence of a pyridine trans to the oxo ligand in C diminishes both the stability and the reactivity of the system. The mechanism of oxidative chlorination catalyzed by [Fe(TPA)(Cl)2]+ was theoretically studied by Siegbahn.454 The reaction of this compound with a peroxide yields [Fe(OOR)(TPA)(Cl)]+, which generates either [FeV(O)(TPA)(Cl)]2+ by OsO heterolysis or [FeIV(O)(TPA)(Cl)]+, by OsO homolysis (Figure 99). These iron-oxo species are biomimetic complexes of the non-heme halogenase enzymes. The oxidative chlorination of cyclohexane by [FeV(O)(TPA)(Cl)]2+ and [FeIV(O)(TPA)(Cl)]+ was studied considering different spin states. In the case of [FeV(O)(TPA)(Cl)]2+, cyclohexane is initially oxidized yielding a cyclohexyl radical cation. In the subsequent step, the transition state for H abstraction has an unusually long O · · · H distance (1.88 Å) and involves an energy barrier of 12.8 kcal mol-1. The relaxation of this transition state leads directly to the formation of the chlorinated product, by transfer of a Clion. This prevents hydroxylation and explains the exclusive and stereospecific chlorination reaction observed experimentally. In contrast, the [FeIV(O)(TPA)(Cl)]+ species follows the classical two-step rebound mechanism, in which the energy barrier for OH rebound is very similar to that of Cl transfer. Hence, for this particular complex, the calculations did not account for the experimental observations. FeIVdO is generally accepted as the reactive group in CsH oxidation by non-heme iron biomimetics. Nevertheless, the nature of the active species has been debated.

The hydroxylation of cyclohexane by high-valent iron-oxo bispidine complexes was studied by Comba using a combined experimental/theoretical approach.455 The calculations show that the active species may be either an FeIVdO or an FeVdO complex, since both groups promote CsH oxidation by an oxygen rebound mechanism. The FeIVdO species is thermodynamically favored by ca. 12 kcal mol-1, whereas the FeVdO species involves energy barriers that are lower than those associated with FeIVdO by ca. 6 kcal mol-1. These results suggest that the hydroxylation of cyclohexane is promoted by both FeIVdO and FeVdO species in competing parallel reaction pathways. This scenario gives indeed the best fit with the experiments. Some biomimetic FeIVdO complexes are able to promote oxidative CsH addition not only by hydroxylation but also by dehydrogenation. For instance, dihydroanthracene is oxidized to anthracene by [Fe(O)(TMC)]2+. This reaction was theoretically studied at the DFT level by Johansson and Siegbahn.456 Previous NMR studies by Rhode and Que457 revealed that several isomers of the active species are present in the reaction mixture, which contains the counteranion, CF3CO2-, and the solvent, acetonitrile. The coordination environment of iron is different in each of these isomers. The axial ligand trans to the oxo group can be either acetonitrile or the counteranion. In addition, the complex may be cationic or neutral depending on whether an ion pair is formed in solution. This issue was considered in the theoretical study by using five models with different axial ligands and counteranion arrangements. The results show that neither the nature of the axial ligand nor the number of counteranions in the ion pairs have a great effect on the reactivity of [Fe(O)(TMC)]2+. The reaction follows a twostep mechanism consisting of two H abstraction steps (Figure 100). The first step is rate-limiting, with an energy barrier of ca. 16 kcal mol-1, in good agreement with the experimental value of ca. 19 kcal mol-1. The second H abstraction has a much lower energy barrier, mainly of entropic origin. As for CsH hydroxylation, the reaction obeys a TSR regime by starting in the triplet ground state and ending in the quintet state, after spin crossover during the first step. The overall reaction is strongly exothermic, by more than 60 kcal mol-1, due to the formation of the aromatic anthracene product.

5.2.3. Non-heme Iron Enzymes The oxidation of propene by the non-heme taurine/Rketoglutarate dioxygenase (TauD) enzyme was theoretically

806 Chemical Reviews, 2010, Vol. 110, No. 2

Balcells et al.

Figure 101. Reaction mechanism for the oxidation of benzene by [HOsFeIVdO] and [H2OsFeIVdO]+ species.

Figure 100. Dihydroanthracene oxidation by [Fe(O)(TMC)(X)]2+ (X ) MeCN or CF3CO2-).

studied by de Visser.458,459 The FeIVdO center of this enzyme is capable of oxidizing both the CH3 (hydroxylation) and the CHdCH2 (epoxidation) groups of the substrate. Three different spin states, triplet, quintet, and septet, are able to promote the hydroxylation and epoxidation of propene, but the reactivity is controlled by the quintet, which provides the lowest energy pathways. The hydroxylation reaction, with a highest energy barrier of 5.4 kcal mol-1, follows the classical oxygen rebound mechanism and competes with the epoxidation reaction, which has almost the same maximum barrier, 4.8 kcal mol-1. Heme enzymes, such as cytochrome P450 (CP450), were compared with the non-heme TauD enzyme, which is fundamentally different in many ways: Fe is five-coordinated in TauD and six-coordinated in the heme enzymes; the reactivity follows a single-state reactivity (SSR) model in TauD and a two-state reactivity (TSR) model in the heme enzymes; in TauD, both electrons of the oxidized substrate are absorbed by Fe, whereas in the heme enzymes, one electron is taken by Fe and the other by the porphyrin ligand; and finally, the valence molecular orbitals are different in each case. In addition to these differences, the calculations showed that the oxidative power of TauD is much higher than that of the heme enzymes, due to lower energy barriers. The higher reactivity of TauD is assigned to the higher radical character of the reactive Fe center, the stronger FeOsH bond formed in the rate-determining H abstraction step, and the higher exchange stabilization of the key metal FMOs. In a subsequent study, the whole TauD enzyme was modeled at the QM/MM level.460 These calculations showed that despite the many H-bond and salt-bridge interactions between the reactive center and the protein environment, the structure and reactivity are essentially the same as those found in the previous DFT models.458,459 Aromatic hydroxylation by tetrahydrobiopterin-dependent hydroxylases was modeled by Bassan at the DFT(B3LYP) level.461 In a previous theoretical study, the authors had demonstrated that the active species, an FeIVdO complex, is generated from molecular oxygen with the assistance of the tetrahydrobiopterin cofactor.462 The oxidation of the aromatic ring of phenylalanine, modeled as benzene, and tryptophan, modeled as indole, was explored considering two different protonation states of the catalytic core:

[HOsFeIVdO] and [H2OsFeIVdO]+ (Figure 101). The oxo and the OH/H2O ligands are bound to the iron(IV) center in a cis position within an octahedral environment. The other four coordination sites are occupied by a monodentate glutamate and an aqua ligand in cis and by two cis histidine molecules. The solvent is explicitly considered in the model by introducing two molecules of water; one to stabilize the octahedral coordination of iron and the other to better model the proton transfers involved in the reaction. With the neutral model, [HOsFeIVdO], the oxidation of the aromatic substrate, in which the FeOsPh bond is formed, involves a phenyl radical. A subsequent 1,2-hydride shift leads to the formation of a keto intermediate, which by keto-enol tautomerization yields the final hydroxylated product. An alternative final step with a higher energy profile, involving the formation of an epoxide, was discarded. With the cationic model, [H2OsFeIVdO]+, the formation of the FeOsPh bond involves a 2e- oxidation yielding an arenium cation. For both catalysts and substrates, the higher energy barrier corresponds to the oxidative formation of the FeOsPh bond. In the case of benzene, this step involves an energy barrier of 16.2 kcal mol-1 for the radical mechanism ([HOsFeIVdO]) and of 10.0 kcal mol-1 for the cationic mechanism ([H2OsFeIVdO]+). Similar barriers were found for indole. The critical energy barrier is lowered by the protonation of the OH ligand. The reaction occurs on the quintet surface, which is also the ground state of the isolated catalyst. The hydroxylation of the benzylic carbon of the substrates was also investigated, considering the oxidation of toluene by the rebound mechanism. In this case, H abstraction has the highest energy barrier, 11.1 kcal mol-1. The results obtained in this study showed that an FeIVdO center in the catalytic core of hydroxylases is capable not only of aromatic hydroxylation but also of benzylic hydroxylation, in good agreement with experiments. A similar catalytic system in which L-phenylalanine is oxidized to L-tyrosine by a non-heme iron hydroxylase was studied by Yoshizawa with QM and QM/MM methods.463 The hydroxylation of the substrate by the active site of the enzyme, which contains an FeIVO active species, may involve two different mechanisms: (1) the classical oxygen rebound mechanism, with CsH cleavage by H abstraction followed by OH rebound or (2) oxygen insertion by electrophilic aromatic addition, followed by 1,2-hydride shift, which yields the keto tautomer of the reaction product, as in Figure 101. The rebound mechanism was explored by considering the triplet, quintet, and septet states of the system. The calculations show that the energy barriers associated with the H abstraction and OH rebound steps amount to 24.6 and 9.2 kcal mol-1, respectively. In contrast, in the oxygen insertion mechanism, the electrophilic addition is barrier-free and the

CsH Bond Activation in Transition Metal Species

Figure 102. Reaction mechanism for the oxidative cyclization of proclavaminic acid (R ) CH2CH2NH3+).

∆Eq for the 1,2-hydride shift is only 9.0 kcal mol-1. This study thus showed that the insertion mechanism is favored over the rebound one. The oxidative cyclization of proclavaminic acid catalyzed by clavamic acid synthase was theoretically studied by Borowski and Siegbahn.464 A realistic structure of the enzyme-substrate complex was obtained by molecular dynamics simulations. The reaction mechanism was explored by means of DFT(B3LYP) calculations on a large model of the reactive center, which contains an FeIVO active species. A new reaction mechanism was proposed on the basis of the calculations (Figure 102): (1) oxidation of the hydroxyl group by reduction of the metal center to FeIIIOH, (2) retroaldol-like decomposition of the O radical intermediate, (3) CsH cleavage by H abstraction, in which the metal center is further reduced to FeIIOH2, and (4) 1,3-dipolar cycloaddition yielding the final reaction product. The CsH cleavage step involves an energy barrier of 20.6 kcal mol-1. Interestingly, the calculations show that this mechanism involves a reaction pathway lower in energy, and thus more favorable, than the mechanism originally proposed for this reaction,465 which consists of radical H abstraction followed by transfer of one electron and one proton. In addition, the new mechanism is consistent with experimental isotope kinetics data.466 Another cyclization reaction of high relevance is the biosynthesis of four-membered β-lactam rings, which are present in several antibiotic drugs, by isopenicillin N synthase. This reaction has been studied by Siegbahn at the DFT level.467,468 The calculations showed that the mechanism involves two key steps: (1) cysteine CsH activation by a radical Fe(III) superoxo species, which has an energy barrier of 14.6 kcal mol-1, and (2) valine CsH activation by a ferryl Fe(IV) oxo species, which has an energy barrier of 15.3 kcal mol-1. These energy barriers are the highest found along the reaction pathway, which suggests that both steps are partially rate-determining, in good agreement with the experiments. Interestingly, the order in which the CsH bonds are cleaved is determined by the oxidative strength of the active species, since the strong CsH bond of valine can only be cleaved by the ferryl-oxo species. The R-ketoglutarate Fe(II)-dependent AlkB family of enzymes remove the alkylation damage from DNA nucleobases by oxidative dealkylation. The regeneration of 1methyladenine (AdeCH3+) by the FeIVdO reactive center of these enzymes was theoretically studied by Gauld at the DFT(B3LYP) level.469 The damaging methyl group is oxidized by the FeIVdO moiety yielding AdeCH2OH+. This process involves radical H abstraction, followed by OH rebound (Figure 103). In the next step, the hydroxyl group

Chemical Reviews, 2010, Vol. 110, No. 2 807

Figure 103. Reaction mechanism for the oxidative dealkylation of 1-methyladenine (AdeCH3+) by the AlkB family of enzymes.

Figure 104. Manganese-oxo bond lengths and spin densities (in parentheses) for [Mn(O)(Cl)(por)].

of AdeCH2OH+ is deprotonated by a Bronsted base, which leads to formaldehyde and the repaired base, adenine, by CsN cleavage. This reaction mechanism was studied by exploring the potential energy surfaces associated with the triplet, quintet, and septet states. The calculations showed that H abstraction is the rate-determining step, with an energy barrier of ca. 22 kcal mol-1, in good agreement with the experimental value, ca. 21 kcal mol-1.

5.2.4. Manganese Porphyrin Biomimetics The oxidation of toluene to benzyl alcohol by [Mn(O)(por)(Cl)] (por ) porphyrin) was studied by Eisenstein et al. at the DFT(BP86) level.320 Manganese porphyrins are efficient synthetic catalysts and biomimetic models of the heme cofactors of the cytochrome enzymes. The calculations showed that [Mn(O)(por)(Cl)] has three accessible spin states: a singlet, a triplet, and a quintet, all within an energy range of ∼10 kcal mol-1. As the total spin of the system increases, the MndO distance elongates and the radical character on oxygen (O• character) becomes higher, due to the occupation of antibonding π*(MndO) orbitals (Figure 104). The singlet, which is the ground state, lacks O• character and does not promote CsH oxidation. In contrast, both the triplet and quintet states, having O• character, promote CsH oxidation. In the triplet state, the highest energy barrier, 2.6 kcal mol-1, corresponds to the H abstraction step. The resulting hydroxo intermediate, [Mn(OH)(por)(Cl)], undergoes OH rebound with almost no barrier. The quintet surface is higher in energy in the H abstraction step but lower in the OH rebound, which suggested that triplet f quintet spin-crossover may happen during the reaction. This study showed that the rebound mechanism for Mn(porphyrin)-catalyzed CsH hydroxylation is only favored for spin states with radical character on oxygen. In a subsequent study by the same authors, the influence of the axial ligand, X, on CsH oxidation by [Mn(O)(por)(X)]+ (X ) OH2, OH-, O2-) was rationalized by means of DFT(BP86) calculations.321 The relative concentration of these three prototropic species is determined by the pH. At low pH, [Mn(O)(por)(OH2)]+ is the dominant species, whereas at high pH, [Mn(O)2(por)]- is the dominant species.

808 Chemical Reviews, 2010, Vol. 110, No. 2

Balcells et al.

Figure 106. Superoxo (left), hydroperoxo (center), and oxo (right) forms of the CuB site in DBM.

barriers: 15.6 kcal mol-1 for the H abstraction step and 13.3 kcal mol-1 for the OH rebound step.

5.2.6. Copper Monooxygenases Figure 105. Energy differences, in kcal mol-1, between the singlet (S), triplet (T), and quintet (Q) states of [Mn(O)(X)(por)]+ (X ) OH2, OH-, O2-).

Although the trans influence of X increases from X ) OH2 to X ) O2-, causing the elongation of the MndO bond, the reactivity in CsH oxidation decreases. The theoretical study showed that the O• character is essential for efficient reaction, as previously shown for [Mn(O)(por)(Cl)].320 For all X, the singlet ground state, lacking O• character, is nonreactive, whereas the high-spin states, triplet and quintet, having O• character, promote CsH oxidation. The low-spin/high-spin energy difference, ∆ELS/HS, is determined by the nature of the axial ligand.470 As the trans influence of X is reduced by making the pH more acidic, ∆ELS/HS decreases, and the reactive high-spin states become thus more accessible, which promotes the reaction (Figure 105). This is due to the energy of the singly occupied π*(MndO) orbitals in the high-spin states, which becomes lower as the π-donor character and trans influence of the axial ligand decreases. This study showed that CsH oxidation by [MnO(por)(X)]+ is promoted by changing the pH from high, X ) O2-, to low, X ) OH2, in total agreement with experiments.

5.2.5. Copper Biomimetics CuI R-ketocarboxylate complexes with supporting N-donor ligands are biomimetics of copper-containing oxygenase enzymes. The reactivity of these complexes was studied at the DFT and CASSCF/CASPT2 levels by Gagliardi, Tolman, and Cramer.471 The oxidation of the CuI complex by O2 may yield a mixture of Cu-peracid and Cu-oxo species, which are both capable of hydroxylating a phenyl ring of the supporting ligand. The calculations show that the oxo species are more reactive than the peracid species. Formally, the most reactive complex contains the CuIIIdO oxo group. Nevertheless, the study revealed that this group is better described as a CuIIsO• oxyl. The introduction of different substituents at several positions of the metal backbone showed that an electron-donating ligand increases the Cu-O bond strength. This makes the Cu-oxyl species more stable than the peracid species but also reduces its reactivity. The triplet state of the Cu-oxyl species promotes aromatic hydroxylation, with the following two-step reaction mechanism: (1) O-attack to the ortho C of the phenyl ring and (2) H transfer from the ipso carbon to the oxygen. This mechanism is analogous to that described by Borowski and Siegbahn for [Fe(TpPh2)BF] (Figure 98).450 The highest activation barrier associated with this process is only 10.1 kcal mol-1. Interestingly, the reaction may also follow the oxygen rebound mechanism. Nevertheless, this mechanism involves higher activation

The catalytic mechanism of dopamine β-monooxygenase (DBM) was explored computationally by Yoshizawa in order to identify the active species and reveal the function of the surrounding amino acid residues.472 DBM contains two reactive copper sites, CuA and CuB, which promote the stereoselective hydroxylation of dopamine. A 3D structure of the whole enzyme was constructed by homology modeling from the crystal structure of a closely related enzyme, peptidylglycine R-hydroxylating monooxygenase. This structure, containing about 4700 atoms, was optimized at the QM(B3LYP)/MM(Amber96) level. The calculations showed that copper is coordinated to three histidine residues in the CuA site and to two histidines and one methionine in the CuB site. Three other residues, two glutamates and one tyrosine, play a key role in orienting the substrate through H bonds to promote H abstraction in a stereospecific fashion. The reactivity of the active center was modeled with DFT(B3LYP) calculations on small models extracted from the QM part of the QM/MM geometries. Three active species, with different charges and spin states, were investigated: Cu-superoxo, Cu-hydroperoxo, and Cu-oxo (Figure 106). The Cu-superoxo species is able to promote H abstraction with an energy barrier of 16.9 kcal mol-1 but is less reactive than the Cu-oxo, which involves a barrier of only 3.8 kcal mol-1. The Cu-hydroperoxo species was ruled out because it is associated with an energy barrier of ∼40 kcal mol-1, which is too high in physiological conditions. In a subsequent study by the same authors, the reactivity of the copper center is studied on the whole enzymatic system by QM/MM calculations.473 In this extended model, the energy barrier for the H abstraction step was 23.1 kcal mol-1 for the Cu-superoxo species and 5.4 kcal mol-1 for the Cu-oxo. The energy of the hydroxo radical intermediate was found to be 18.4 kcal mol-1 in the Cu-superoxo pathway and -14.2 kcal mol-1 in the Cu-oxo. These results indicate that the active species is likely to contain a CudO oxo group, which according to the calculations is better described as a CusO• oxyl group. The protein environment favors the hydroxylation reaction both kinetically and thermodynamically. All steps are clearly exothermic and involve low energy barriers, supporting the oxygen rebound mechanism. The catalytic mechanism of peptidylglycine R-hydroxylating monooxygenase was studied by Amzel and Estrin.474 This copper-containing enzyme catalyzes the R-hydroxylation of terminal glycine residues of proteins (Figure 107). The full enzymatic system solvated by water, with a total of 23 441 atoms, was modeled with a QM/MM approach. The reactive CuB site, included in the QM part, contains a copper atom coordinated to the oxidant, O2, the substrate, formyl glycine, and the residues of the protein side chains, two histidines and one methionine. The calculations show that

CsH Bond Activation in Transition Metal Species

Figure 107. Oxidation of terminal glycine residues of proteins by R-hydroxylating monooxygenase.

O2 binds to the CuB site yielding a CuIIsO2•- species. The binding of O2 is promoted by previous coordination of the substrate. The reduction of this compound to CuIsO2•- by electron transfer from the CuA site increases the stability of the system due to higher π-back-donation. Neither of these two species is capable of H abstraction. Spontaneous proton transfer from the solvent generates a CuIIsOOH+ compound, which is not active in H abstraction either. Nevertheless, a second proton transfer yields a CudO2+ complex by water release. In the quartet ground state of this species, two unpaired electrons localized in a CuIIIdO+ moiety, are ferromagnetically coupled with a radical cation delocalized over the protein residues coordinated to the metal. The CudO2+ complex promotes the hydroxylation of the substrate in a concerted manner with almost no energy barrier. The reduction of CudO2+ by ascorbate yields CudO+, which is also active in H abstraction but with a classical two-step rebound mechanism involving higher energy barriers. Both species undergo spin inversion during the reaction. In view of the results obtained, the CudO2+ complex was proposed as the active species. This study revealed that the catalytic mechanism of peptidylglycine R-hydroxylating monooxygenase is analogous to that of cytochrome P450.

5.2.7. Dehydrogenases The mechanism of hydrocarbon oxidation by ethylbenzene dehydrogenase was studied by Szaleniec at a DFT level.475 This enzyme, which contains a molybdenum oxo group in its reactive center, catalyzes the stereoselective oxidation of ethylbenzene to (S)-1-phenylethanol. Three different mechanisms were considered, depending on the cationic or radical

Chemical Reviews, 2010, Vol. 110, No. 2 809

character of the intermediates (Figure 108). In mechanism I, the substrate undergoes homolytic CsH cleavage yielding a carbocation and a Mo(IV)sOH intermediate. In a subsequent step, the carbocation attacks the hydroxo ligand yielding the final reaction product. Mechanism II is the classical oxygen rebound mechanism. Mechanism III is a combination of mechanisms I and II, in which homolytic CsH cleavage is followed by electron transfer leading to the carbocation and the Mo(IV)sOH intermediate. A simple quantitative structure-activity relationship (QSAR) model was developed in order to determine the nature of the reaction mechanism. This model suggested that the transition state has a predominant carbocationic character. Mechanism I is thus preferred, but if the rate is controlled by the OH-rebound step, mechanism III cannot be discarded.

5.3. Metal-Hydroxo Species 5.3.1. Lipoxygenases The role of nuclear quantum mechanical (NQM) effects in enzyme catalysis was explored in the case of lipoxygenases by Siegbahn and Warshel.476 In biological systems, lipids are catalytically oxidized by lipoxygenases. The reaction involves radical H abstraction from the aliphatic chain of the lipid by an FeIIIsOH active species, yielding a carbon radical (Figure 109). This is the critical step of the mechanism, as indicated by the extremely large kinetic isotopic effects (KIE) observed experimentally. The reaction was studied by applying the quantum classical path (QCP) approach in order to model the NQM effects, which due to the H transfer nature of the reaction include large tunneling effects. The experimental KIE was well reproduced, as well as the temperature dependence of the H abstraction rate constants. In the case of D abstraction, the simulation was less successful. The calculations showed that the influence of the media, either the enzyme or a solution, is essentially irrelevant. The outer-sphere reorganization energies were found to be extremely low, due to the very small charge

Figure 108. Oxidation reaction mechanisms proposed for ethylbenzene dehydrogenase.

810 Chemical Reviews, 2010, Vol. 110, No. 2

Balcells et al.

Figure 109. Reaction mechanism for the oxidation of fatty acids by the FeIII(OH) center of lypoxygenases.

Figure 111. Energy profile, in kcal mol-1, for H abstraction from toluene by [Mn(OH)2(por)]+ in the triplet state.

Figure 110. Model of the lipoxygenases reactive center.

transfer associated with the reaction. Interestingly, the NQM effects were found to be the same in the enzyme and in solution, which suggests that they do not contribute to the catalysis. This study proved that the QCP approach can be successfully applied to model the contribution of NQM effects to activation free energies in the framework of enzymatic catalysis, even when large tunneling effects are involved. Proton-coupled electron transfer (PCET) in soybean lipoxygenases was theoretically studied by HammesSchiffer.477 The first step in lipid oxidation by lipoxygenases (Figure 109) can be regarded as a PCET, a process in which one proton and one electron, that is, a H•, are transferred simultaneously. The reaction was studied by applying the multistate continuum theory, in which the transferred hydrogen is modeled with a quantum mechanical wave function. The reorganization energy associated with the reaction was evaluated with DFT calculations for the iron cofactor (Figure 110) and with a frequency-resolved cavity model for conformations obtained with docking calculations. The values computed for these energies were ∼19 and ∼2 kcal mol-1, respectively. The precited dependence of the rate constant and KIE on the temperature was in good agreement with experiments, which supported the PCET mechanism. The dependence of the rate constants on the temperature is low due to the small ∆Gq of the reaction, which results from the balance between the total reorganization energy and the ∆G. The high deuterium KIE observed for this reaction, 81, is due to the following factors: (1) the small overlap between the proton vibrational wave functions of the reactant and the product and (2) the dominance of the lowest vibronic states of the reactant and the product for the tunneling process. This study reveals that the proton donor-acceptor vibrational motion plays a key role in reducing the donor-acceptor distance relative to its equilibrium value, which facilitates the PCET process. In a subsequent study, the dynamics of this enzymatic reaction was explored by the same authors.478 The reaction

was investigated with a vibronically nonadiabatic approach, in which the active electrons and the transferred proton were treated at a quantum mechanical level. The motion of all atoms in the full solvated enzyme system was also included in the model. The calculations showed that the dynamic behavior of the system is dominated by the motion of the protein and the solvent and not by the proton donor-acceptor motion. The rate constant was calculated as the time integral of a probability flux correlation function. The magnitude of the overall rate is strongly influenced by three different factors: (1) the proton donor-acceptor frequency, (2) the vibronic coupling, and (3) the protein/solvent reorganization energy, which was estimated to be ∼39 kcal mol-1.

5.3.2. Manganese Porphyrin Biomimetics CsH oxidation by hydroxo manganese(V) porphyrins was studied by Eisenstein at the DFT(BP86) level.322 These complexes mimic the catalytic properties of cytochrome P450 enzymes. The rebound mechanism was explored considering toluene as model substrate and porphin as model porphyrin. In the classical rebound mechanism, radical H abstraction is promoted by the MnVO group, but in this study, the authors postulate MnVOH as the active group. In the acidic conditions required for reaction, two different prototropic species can be present: [Mn(O)(por)(OH)] and [Mn(OH)2(por)]+ (por ) porphyrin); both species contain the MnVOH group. The ground state of [Mn(OH)2(por)]+ is the triplet, followed by the quintet, 0.3 kcal mol-1 above, and the singlet, 16.2 kcal mol-1 above. The triplet and quintet states promote toluene hydroxylation by the rebound mechanism, involving exothermic steps with low energy barriers. The highest energy point along the reaction pathway is only 0.9 kcal mol-1 above reactants (Figure 111). The aqua tautomer of the dihydroxo complex, [Mn(O)(por)(OH2)]+, which can only react with the MnVO group, requires a maximum energy gain of 4.8 kcal mol-1. These results suggested that MnVOH is capable of promoting CsH oxidation. In contrast, in the neutral [Mn(O)(por)(OH)] complex, CsH oxidation can be promoted by both axial ligands, but preferentially by MnVO. This study showed that more acidic conditions not only make the Mn(por) catalysts more reactive but also change the nature of the reactive center from MnVO to MnVOH. CsH

CsH Bond Activation in Transition Metal Species

Chemical Reviews, 2010, Vol. 110, No. 2 811

Figure 112. Transition state for the CsH oxidation of cis-1,2dimethylcyclohexane by [Re(O)(η2-O2)2(CH3)].

oxidation is promoted by MnVO and MnVOH because in both cases there are singly occupied π*(MndO) orbitals. The relevant contribution of oxygen to these orbitals has a double consequence: (1) it increases the electrophilic character of oxygen, which promotes the nucleophilic attack of the CsH bond, and (2) it introduces the oxyl character, which promotes the radical H abstraction from the substrate.

5.4. Metal-Peroxo Species 5.4.1. Methyltrioxorhenium The catalytic CsH oxidation of several substrates, including cis-1,2-dimethylcyclohexane, toluene, 1-phenylethanol, and 1-phenylethyl methyl ether, by methyltrioxorhenium (MTO), [Re(O)3(CH3)], was studied by Privalov at the DFT level.479 In the presence of hydrogen peroxide, MTO yields a mixture of two peroxo complexes: the dioxo peroxo complex [Re(O)2(η2-O2)(CH3)], and the oxo diperoxo complex [Re(O)(η2-O2)2(CH3)]. Both complexes are active in CsH oxidation, but the diperoxo is much more reactive. The calculations showed that the reaction follows a single-step concerted mechanism without the formation of ionic intermediates, even in polar solvents. The oxidation process promoted by the diperoxo complex involves the concomitant cleavage of the CsH bond of the substrate and the OsO bond of the catalyst. In the transition state (Figure 112), the C · · · H · · · O moiety is aligned in a linear manner, resembling the H abstraction step of the oxygen rebound mechanism. The relaxation of this TS, followed by IRC calculations, showed that once the hydride is transferred to the peroxo ligand, the Re-OH bond rotates to undergo oxygen rebound with the resulting carbocation, without the formation of any intermediate. This mechanism has energy barriers of 20-23 kcal mol-1 for the different tested substrates, which are much more reasonable than those of the classical oxidation mechanisms involving metal alkoxide and metal carbonyl intermediates. The predicted KIE, kH/kD ) 2.9, was in very good agreement with the experimental value, 3.2. The butterfly transition states associated with a side-on approach of MTO to the CsH bond, previously suggested by analogy with the oxidation mechanisms of dimethyldioxirane,480,481 were discarded.

5.5. Metal-Hydroperoxo Species 5.5.1. Activated Bleomycin Bleomycin, BLM, is a glycopeptide used as an antibiotic chemotherapy agent. In its active form, ABLM, it is coordinated to a low-spin FeIIIsOOH center. ABLM induces single- and double-strand DNA damage by H abstraction from the C-4′ deoxyribose sugar moiety. The mechanism of this CsH bond cleavage has been debated. Two alternative mechanisms (Figure 113) are possible: (1) a two-step mechanism, starting with the homolysis of the OsO bond, like in the heme chemistry, followed by H abstraction by the OH radical, which leads to a carbon radical and a molecule of water and (2) a direct mechanism, in which the

Figure 113. Stepwise and concerted mechanisms for CsH activation by activated bleomycin.

same reaction products are obtained in a single concerted step from the FeIIIsOOH species. A mixed spectroscopic/ computational (DFT) study by Que and Solomon482 showed that OsO bond homolysis in ABLM is feasible, with a free energy gain of 12 kcal mol-1 and no energy barrier, like in the non-heme FeIII complexes [Fe(OOH)(N4Py)]2+ and [Fe(OOH)(TPA)(OH2)]2+. Other theoretical studies by Shaik pointed out that the OsO bond may also undergo heterolytic cleavage after protonation.483 Nevertheless, previous experiments revealed the presence of prominent KIEs, kH/kD ≈ 2-7,484 which are too large for H abstraction by OH•, thus supporting the direct H abstraction mechanism. The structure and reactivity of ABLM was studied by Solomon using a combination of spectroscopic and computational (DFT) techniques.485,486 The calculations showed that ABLM promotes H abstraction without requiring the previous cleavage of the OsO bond, thus following the direct reaction mechanism. In the concerted transition state, the cleavage of the sugar CsH bond is concomitant with the formation of an OsH bond and the cleavage of the FeIIIOsOH bond. The reaction is favorable at the thermodynamic and kinetic levels, with a ∆G of -10 kcal mol-1 and an activation barrier in solution of 17 kcal mol-1. This energy pathway is below those associated with the homolytic and heterolytic cleavage of the OsO bond. Interestingly, the FeIVdO product of the reaction is a reactive species able to promote a second H abstraction, which may explain why a single molecule of ABLM is capable of cleaving both DNA strands. The direct H abstraction mechanism was further supported in a recent study in which the kinetics of the ABLM + DNA reaction was compared with that of the ABLM decay.487 The spectroscopic and DFT results show that the decay reaction, in which ABLM undergoes the intramolecular cleavage of a NsH bond, is slower than the reaction with DNA.

5.6. Bimetallic µ-Oxo Species 5.6.1. Copper Biomimetics The mechanism of intramolecular CsH activation in [(LCu)2(µ-O)2]2+, L ) 1,4,7-trialkyl-1,4,7-triazacyclononane, was theoretically studied by Cramer with a QM/MM approach.488 [(LCu)2(µ-O)2]2+ is a biomimetic complex of copper-containing metalloenzymes. Experiments had shown that this compound undergoes N-dealkylation yielding benzaldehyde. The calculations showed that this reaction involves intramolecular CsH activation by a classical rebound mechanism. In the initial rate-determining step, a benzylic H is transferred to a bridging oxo group. The transferred H is on the equatorial plane defined by the diamond Cu2O2 core. The reaction yields a copper hydroxo intermediate, by breaking the Cu-O-Cu bridge. This intermediate has a mixed radical/ionic character. In a subsequent almost barrierless rebound step, the hydroxo group is transferred

812 Chemical Reviews, 2010, Vol. 110, No. 2

Balcells et al.

Figure 115. Nonradical mechanism proposed for methane oxidation by FeO+.

Figure 114. Reaction mechanism proposed for methane hydroxylation by MMO.

from copper to the oxidized benzylic carbon, leading to the formation of an aminal intermediate, which yields the final benzaldehyde product by hydrolysis. The reaction is weakly sensitive to aromatic substitution in the ligand, due to some polar character in the H abstraction transition state. The calculations also reveal that at 233 K, tunneling plays a significant role in the kinetics of the reaction. This study suggested that this type of copper biomimetic may be potentially active not only in the intramolecular but also in the intermolecular CsH activation.

5.6.2. Methane Monooxygenase Methane monooxygenase (MMO), an enzyme found in methanotroph bacteria, catalyzes the oxidation of methane to methanol, by using O2 as oxidant.216,217 This enzyme has two different forms: cytoplasmic soluble MMO (sMMO) and membrane-bound particulate (pMMO). The sMMO form contains iron in the active site, whereas the pMMO form, which is much more difficult to isolate, purify, and characterize, contains copper. The sMMO form contains three components: hydroxylase (MMOH), B component (MMOB), and reductase (MMOR). The MMOH component binds oxygen and catalyzes the oxidation of methane. The resting state of MMOH is the oxidized form, MMOHox, which contains a diferric FeIIIsFeIII core (Figure 114). After a 2ereduction by NAD(P)H and oxidation by O2 a metastable intermediate with a bridging peroxo ligand, known as intermediate P, is formed. This species leads to an FeIVs (µ-O)2sFeIV diamond core species, known as intermediate Q, which is the active species in methane oxidation. In intermediate Q, the iron atoms are bridged by two oxo ligands and one bidentate glutamate ligand. In addition, each Fe is bound to other glutamate and histidine residues in terminal positions and a molecule of water may be also present within the first coordination sphere. The exact structure of the active site of intermediate Q is not fully clear, and different computational models have been proposed in order to study the mechanism of methane hydroxylation by MMO. In a first approach, the enzyme was modeled by means of DFT(B3LYP) calculations on small model complexes of the reactive site. Four different models were proposed by the research groups of Yoshizawa,489–491 Morokuma-Basch,492 Siegbahn,493–497 and Friesner-

Lippard.498–501 These models differ in their size, spin state, charge, and Fe coordination number. Different basis sets were used in each case to optimize the geometries and refine the energies, and the calculations were carried out using different computational programs. In a second generation of more sophisticated studies, the whole enzyme and surrounding molecules of solvent were included in the model, by means of hybrid QM/MM calculations. Remarkably, despite the diversity of models and methods, all theoretical studies indicated that H abstraction from methane by the bridging µ-oxo ligand is the rate-determining step of the process. The reaction mechanism of CsH activation by bare metal-oxo species in the gas phase has been extensively studied by means of spectroscopic and computational techniques by Schro¨der, Schwarz, and Shaik.502–505 In line with these studies, FeO+ was initially considered by Yoshizawa as the simplest possible model of sMMO. In this model, the formal oxidation state of Fe is not IV, as it is in intermediate Q, but III instead, and the reaction is not catalytic. Nevertheless, experiments have shown that this bare metal-oxo cation oxidizes methane to methanol under ion cyclotron resonance conditions.506,507 The hydroxylation mechanism proposed by Yoshizawa508–515 is represented in Figure 115. In the starting intermediate, methane is bound to FeO+ through two weak Fe · · · H bonds. From this species, the activation of the CsH bond involves a four-center transition state leading to a hydroxo methyl iron intermediate. In a final reductive elimination step, referred to as “methyl migration”, the concerted cleavage of the Fe-C bond and formation of the CsO bond, yields a molecule of methanol and an FeI cation. This model of intermediate Q was subsequently extended by Yoshizawa considering two different diiron complexes, one with aqua terminal ligands516,517 and the other with hydroxo terminal ligands,518 and in both cases with one glutamate and two oxo bridging ligands. The complex electronic structure of the system was initially modeled assuming a closed shell singlet configuration with all spins paired. The rate-determining step of the mechanism is the migration of one hydrogen from methane, which is initially coordinated to the unsaturated Fe atom, to one of the bridging oxo moieties (Figure 116). The relaxation of the transition state, 31 kcal mol-1 above reactants, leads to an intermediate with a µ-hydroxo bridge, in which the resulting methyl fragment is bound to the metal. In the next step, the methyl ligand migrates from the metal to the µ-hydroxo bridge yielding a methanol complex, 39 kcal mol-1 below reactants. In this mechanism, which does not have radical character, the formal oxidation state of the metal does not change in the H abstraction step. The two-electron oxidation of methane is associated with the methyl migration step. A subsequent study519 on the antiferromagnetic state by broken-symmetry DFT calculations520 showed that the magnetic coupling

CsH Bond Activation in Transition Metal Species

Chemical Reviews, 2010, Vol. 110, No. 2 813

Figure 116. Nonradical mechanism proposed in the Yoshizawa model of sMMO.

Figure 117. Mononuclear and dinuclear models of the copper active sites of pMMO.

between the two metal centers may shift the whole pathway down in energy, leading to lower energy barriers in better agreement with experiments. The mechanism of methane hydroxylation by sMMO was recently further explored by Yoshizawa using a more sophisticated model.521 In this model, the two metal centers are bridged by two µ-oxo ligands and a bidentate glutamate carboxylate, modeled as formate. One of the iron atoms is coordinatively saturated with three extra ligands: two monodentate formates and one imidazole ring, used as a model of histidine. The other iron atom is coordinated with two extra ligands, one formate and one imidazole, and has thus a vacant coordination site, as supported by experimental EXAFS data.522 The calculations were carried out for the broken-symmetry singlet state, which involves two high-spin FeIV centers antiferromagnetically coupled. The iron vacant site is occupied by methane, which undergoes heterolytic CsH cleavage promoted by one of the µ-oxo ligands. In the subsequent step, the methyl ligand is transferred from the metal to the µ-hydroxo ligand through a three-centered transition state. The CsH cleavage is the most difficult step of the mechanism, with a high energy barrier of 36.5 kcal mol-1. In the context of the classical oxygen rebound mechanism, a transition state for radical H abstraction by the µ-oxo ligand, without the direct participation of Fe, was also optimized. Nevertheless, this stationary point could only be located in the high-spin nonet and undecet states and not in the antiferromagnetic singlet ground state. Changing the electronic nature of the CsH cleavage from heterolytic to homolytic does not alter to a great extent the kinetic isotopic effects associated with this step. This study supported a two-step mechanism, in which the substrate does not generate any radical or ionic species. This mechanism is essentially identical to that of the simpler FeO+ and diiron systems. The oxidation of methane by the particulate form of methane monooxygenase (pMMO) was also explored by Yoshizawa, combining DFT and QM/MM methods.523 Both the monocopper and dicopper active sites of the enzyme were modeled (Figure 117). The oxidation of the mononuclear site by O2, leading to an oxo CuIII species, is endothermic. In

contrast, the oxidation of the dinuclear site, leading to a bis(µoxo) CuIICuIII species, is exothermic. Both active species promote the oxidation of methane under physiological conditions. The reaction mechanism is analogous to that of sMMO and involves H abstraction by the terminal (mononuclear site) or bridging (dinuclear site) oxo ligand, yielding a methyl fragment that is stabilized by coordination to copper. The intermediate undergoes methyl migration from the metal to the hydroxo group, giving rise to methanol. The overall process is strongly exothermic by more than 45 kcal mol-1. Interestingly, the identity of the rate-determining step depends on the nature of the active site. In the mononuclear model, the H abstraction step has the highest energy barrier and may involve two different spin states, singlet and triplet, with similar energies. In the dinuclear model, the methyl migration step has the highest energy barrier and the reactive state is a doublet. In both cases, the σ* orbital of the Cu-O bond is singly occupied and plays a crucial role in the cleavage of the CsH bond. The predicted kinetic isotope effect at 300 K was in reasonable agreement with experiments. In a very recent study, the nature of the active site of pMMO was further studied by comparing the reactivity of two bis(µ-oxo) complexes with different oxidation states in the bimetallic core, CuIICuIII and CuIIICuIII.524 The calculations showed that the bis(µ-oxo) CuIICuIII complex promotes the CsH bond activation of methane, whereas the bis(µoxo) CuIIICuIII complex is nonreactive. Nevertheless, this complex can undergo an intramolecular isomerization process, in which one of the CusO bonds is cleaved with an energy cost of 23.6 kcal mol-1, yielding a CuIIIsOsCuIIIsO species. Interestingly, the terminal oxygen of this species promotes the oxidation of methane to methanol, through the same two-step nonradical mechanism previously proposed by Yoshizawa for the Fe and Cu models of sMMO and pMMO. The bis(µ-oxo) CuIICuIII complex follows the same mechanism without requiring the previous cleavage of a CusO bond. For both complexes, the methyl migration step, in which methanol is formed, is predicted to be slower than the H abstraction step, in which the CsH bond is cleaved. The energy of the migration transition state relative to reactants is 35.8 kcal mol-1 for the bis(µ-oxo) CuIICuIII complex and 53.5 kcal mol-1 for the bis(µ-oxo) CuIIICuIII complex, indicating that the bimetallic CuIICuIII core is more efficient in the hydroxylation of methane than the CuIIICuIII core. The oxygen rebound mechanism was excluded due to the high energy barrier associated with the radical H abstraction step. The reactivity of pMMO has been also explored by Chan,525 by means of DFT calculations on small models of

814 Chemical Reviews, 2010, Vol. 110, No. 2

Figure 118. The oxo-transfer concerted mechanism proposed by Chan for pMMO.

the reactive site. Three different models were considered: a dinuclear bis(µ-oxo) CuIIICuIII complex, a dinuclear bis(µoxo) CuIICuIII complex, and a trinuclear bis(µ3-oxo) CuIICuIICuIII complex proposed by the same authors in a previous study.526 The calculations show that the trinuclear species promotes methane oxidation more efficiently than the dinuclear, with a reaction rate higher by 5-6 orders of magnitude. The kinetic isotope effect predicted for the trinuclear species, 5.2 at 300 K, is in good agreement with experiments. The reaction mechanism, which consists of a direct oxo-transfer process, is different from that proposed by Yoshizawa. Methane is hydroxylated by the insertion of one of the bridging oxo ligands yielding methanol in a single concerted step (Figure 118), with an energy barrier of 15 kcal mol-1. This mechanism is consistent with the predominant retention of configuration observed experimentally, which suggests the absence of radical intermediates. The mechanism of methane oxidation by sMMO was also investigated by Morokuma and Basch.527–531 In order to avoid computational complications, a high-spin configuration with two ferromagnetically coupled FeIV centers was assumed.

Balcells et al.

The model complex has neutral charge, in contrast with the cationic models proposed by Yoshizawa. In the original Morokuma-Basch model, the histidine ligands were modeled as NH2-, the terminal glutamates as H2O, and the bridging glutamates as HCOO-. The main drawback of this simple model is that the charges of the real system, neutral for histidine and anionic for the terminal glutamate are reversed in the model. This was improved in a subsequent extended model,531 in which the histidine residues were modeled as imidazole rings and all the glutamates, both terminal and bridging, as formates. In this model, each iron is coordinatively saturated with a total of six ligands in a nearly octahedral arrangement. This is indeed the main difference with respect to the Yoshizawa models, in which the iron promoting CsH oxidation has a vacant site. The absence of a vacant site causes a major change in the reaction mechanism, which in the Morokuma-Basch model becomes the classical oxygen rebound mechanism (Figure 119). In contrast with the nonradical mechanism proposed by Yoshiwaza, in the rebound mechanism one electron is transferred from methane to the diiron core at each step. The bridging oxo group in the side of the complex occupied by the terminal formate ligands, which has radical character, promotes H abstraction from methane.531 This step involves the transfer of a H radical and can be thus viewed as a proton-coupled electron transfer.532 The transition state has a linear O · · · H · · · C arrangement and leads to a µ-hydroxo intermediate bound to a methyl radical, which due to the absence of vacant sites does not coordinate to iron. In the next step, the carbon radical undergoes OH rebound with the µ-hydroxo ligand, yielding a methanol-bound product intermediate. This species contains an FeIIIFeIII core, which in the ferromagnetic configuration approach is an 11A state, whereas the starting FeIVFeIV species is associated with a 9A state. Both spin surfaces were modeled, and the calculations indicated that the reaction starts on the 9A surface and finishes on the 11A one after spin crossover in the OH rebound transition state region. These results should be taken

Figure 119. The radical oxygen rebound mechanism in the Morokuma-Basch model of sMMO.

CsH Bond Activation in Transition Metal Species

Figure 120. The Siegbahn (top) and Friesner (bottom) models of sMMO, with one of the bridging carboxylate ligands shifted.

with much care though, due to the unrealistic nature of the 9 A and 11A states; Mo¨ssbauer and EPR experiments533–535 showed that both intermediate Q and the final methanol product are singlet species. The radical H abstraction step is rate-determining, with an associated energy barrier of 13 kcal mol-1, ca. twice the height of the OH rebound barrier, 7 kcal mol-1. The calculations showed that the energy barrier associated with the radical H abstraction step decreases as the cleaved CsH bond becomes weaker,531 from HsCH3 to HsCH2F and HsCH2CH3, as expected. Interestingly, the reaction pathway associated with the radical oxygen rebound mechanism of the Morokuma-Basch model is clearly lower in energy than that associated with the nonradical mechanism of the Yoshizawa model. The reactivity of sMMO was also investigated by Siegbahn.536–539 As in the Morokuma-Basch approach, a neutral complex was used to model the reactive center of intermediate Q. In the original model,536 all ligands other than the two µ-oxo bridges were represented with water and hydroxo ligands. It was then found that two bridging complexes should be introduced in the model to reproduce properly the experimental Fe · · · Fe distances.537 The model was further refined by including imidazole and formate ligands to account for the histidine and glutamate residues, respectively, present in the real enzymatic system. In the last and more sophisticated study,539 the model was modified to include the carboxylate shift proposed by Friesner and Lippard,540 in which one of the formate bridging ligands rearranges from the bridging coordination mode to the terminal mode (Figure 120), with an energy loss of 10 kcal mol-1. The resulting vacant site is occupied by a molecule of water. Both the ferromagnetic and the antiferromagnetic broken-symmetry configurations were considered. Siegbahn proposed a new key intermediate for the hydroxylation of methane, labeled as Q′, which is a mixed valence FeIIIFeIV species with an unpaired electron in one of the bridging oxygens. This species has a ferromagnetic µ-O(v)Fe(vvvvv)Fe(vVvv) 9A configuration and is generated from intermediate Q by an intramolecular electron transfer from the µ-O ligand to the diiron core. The Q′ intermediate, which is only 2 kcal mol-1 above intermediate Q, promotes the oxidation of methane. The radical H abstraction step has the

Chemical Reviews, 2010, Vol. 110, No. 2 815

highest energy barrier, 14 kcal mol-1, and yields a µ-OH intermediate bound to a •CH3 radical. This species undergoes OH rebound giving rise to the final methanol product. Despite the carboxylate shift difference between the Morokuma-Basch and Siegbahn models, in both cases the calculations suggested essentially the same oxygen rebound mechanism. Noodleman and co-workers studied the structure of methane monooxygenase, taking into account the electrostatic interactions between the active site and the protein environment.541–543 The results showed that the protein has a relevant impact on the structure and energetics of the active site, suggesting that minimum-sized DFT models are not appropriate. In line with this, Friesner, Lippard, and co-workers developed a new computational model by systematically increasing its size,540 with the final aim of reproducing the X-ray crystal structure of MMOHred.544 The best results were obtained with a large-scale model of about 100 atoms (Figure 120), which succeeded in reproducing the most distinctive features of the active site in MMOHred, including the Fe · · · Fe distance. As the Morokuma-Basch and Siegbahn models, the Friesner-Lippard model is fully saturated, with two six-coordinated Fe centers in a pseudooctahedral environment. The system has a low-spin antiferromagnetic configuration, as supported by experiments,533–535 which is computed by means of the DFT broken-symmetry approach. The mechanism of methane hydroxylation by sMMO was studied using the Friesner-Lippard model.545 The reaction follows a particular oxygen rebound mechanism, involving two parallel reaction pathways. The nucleophilic attack of methane to the µ-oxo ligand in the carboxylate side of the active species, promotes the homolytic cleavage of one of the CsH bonds, with an energy barrier of 18 kcal mol-1. Right after the H abstraction step, the reaction pathway bifurcates and two different intermediates can be formed, depending on how the methyl radical binds to the µ-hydroxo ligand. In the pathway labeled as “bound-radical”, •CH3 binds to µ-OH with an unusually short C · · · O distance of 1.97 Å. In the other pathway, labeled as “nonsynchronous concerted”, this distance is even shorter. These species have similar energies and undergo OH rebound through very low energy barriers, yielding the same methanol complex product in a strongly exothermic fashion. In this mechanistic framework, the authors suggest that •CH3 is indeed a bound radical with a very short lifetime. In addition, the lifetime depends on the pathway followed by the system and is predicted to be significantly shorter in the “nonsynchronous concerted” path, which has an associated rebound barrier of only 1.3 kcal mol-1, than in the “bound-radical” path, which has a higher rebound barrier of 3.9 kcal mol-1. All these results have an intuitive connection with the experiments done on the racemization of CH3CHDT, which revealed that retention of configuration dominates over inversion.546,547 The CH3CHDT racemization experiments were modeled by means of semiclassical molecular dynamics calculations.548 In the “nonsynchronous concerted” pathway, the substrate is oxidized with total retention of configuration. In the parallel “bound-radical” pathway, inversion takes place preferentially due to the lifetime of the ethyl radical, 320 fs, which is significantly longer than the time required to rotate the CsC bond of the substrate. The inversion/retention ratio is indeed determined by the radical lifetime and the rate of the CsC bond rotation. Interestingly, this rate depends strongly on the C · · · O distance between the carbon radical

816 Chemical Reviews, 2010, Vol. 110, No. 2

Balcells et al.

Figure 122. The key hydrogen atom transfer step in the oxidation of alcohols by galactose oxidase.

Figure 121. Electron flow in the oxygen rebound mechanism for the hydroxylation of methane by sMMO.

and the oxygen of the µ-OH ligand; the closer is the radical to the diiron core, the slower is the CsC bond rotation, favoring retention of configuration over inversion. On the basis of the results obtained, the authors predicted a retention of configuration of 69-84%, in agreement with the experimental result, 72%. The rearrangement of the electronic structure along the reaction pathway was analyzed in the Friesner-Lippard model (Figure 121).549 The binding of methane induces a single electron transfer from a lone pair of a bridging oxygen to one of the FeIV centers, which is reduced to FeIII. This process involves the promotion of one electron from the occupied π(Fe2O2) bonding orbitals, consisting of in-phase combinations of the Fe dx2-y2 and the O px/y orbitals to the corresponding π*(Fe2O2) antibonding orbitals. Such electron transfer makes the µ-O ligand radicaloid and more electrophilic, thus promoting H abstraction at this site by absorption of one electron from the σ orbital of the CsH bond. The resulting intermediate undergoes a second electron transfer from a lone pair of the µ-OH ligand to the remaining FeIV center, which is also reduced to FeIII. This yields a radicaloid and electrophilic µ-OH•δ+ moiety, which readily reacts with the •CH3 radical collapsing into the final methanol FeIIIFeIII product complex. The singlet antiferromagnetic configuration of the system is kept all along the reaction pathway, and no spin crossover is thus needed. The Friesner-Lippard model was further refined including the protein environment surrounding the active site, by means of mixed QM/MM calculations.550 The oxidation of several CH3X (X ) H, CH3, OH, CN, and NO2) substrates by sMMO was explored. The calculations were used to rationalize the kinetic data obtained in a previous experimental study.551 Two steps were analyzed in detail, the initial binding of the substrate to the enzyme, which has a nonnegligible diffusion barrier, and the subsequent H abstraction. For X ) CN and NO2, the oxidation reactions were run under saturating conditions, and therefore the activation barrier associated with H abstraction was computed for comparison with experiments. In this case, an excellent agreement with the empirical values was obtained, with deviations of 1-2 kcal mol-1. For the other substrates, X ) H, CH3, and OH, the activation barrier associated with the substrate binding has to be included in the model. For methane, the predicted activation barrier was ∼3 kcal mol-1 higher than the experimental measure. This deviation was also obtained for X ) CH3 and OH, and it is probably due to an inadequate sampling of the configurational space or the harmonic approaches used to compute the entropy or both. The consideration of the protein environment effects (van der Waals and electrostatic interactions) is essential for

the accurate calculation of the absolute free energies. The van der Waals interactions stabilize the transition state, and the inclusion of the electrostatic interactions is needed for a proper modeling of the reaction with the more polar substrates.

5.7. Metal-Alkoxo Species 5.7.1. Galactose Oxidase Galactose oxidase catalyzes the oxidation of primary alcohols to aldehydes by O2. The active site of this enzyme contains a Cu(II) center, which binds to the substrate and four amino acid residues in a square pyramidal geometry.552 The equatorial plane is occupied by two histidines in cis, one tyrosine, and the substrate. The axial position is occupied by a tyrosine. The mechanism postulated for the reaction553 involves the following four steps: (1) proton transfer from the hydroxyl group of the alcohol to the axial tyrosine, (2) H atom transfer from the substrate to the equatorial tyrosine, (3) electron transfer from the substrate to the metal center, which yields the aldehyde product by Cu(II) f Cu(I) reduction, and (4) catalyst recovery by reoxidation of the metal center with O2. The experiments suggest that the H atom transfer (Figure 122), in which a CsH bond of the substrate is cleaved, plays a central role, being the ratedetermining step of the process. This reaction mechanism has been also studied theoretically.554–558 The oxidation of methanol to formaldehyde by galactose oxidase was explored by Himo and co-workers employing two different computational models: a small model, computed at the DFT(B3LYP) level, and a more relatistic model, computed at the IMOMM(B3LYP/MM3) level.559 The calculations show that after the proton transfer step, the substrate undergoes radical H abstraction by the Cu-bound O of the equatorial tyrosine, which has a π-radical in the phenyl ring. Interestingly, the location of this π-radical is displaced from the axial tyrosine to the equatorial tyrosine during the initial proton transfer step. The transition state associated with the H atom transfer is the highest energy point along the reaction pathway, which suggests that this step is rate-determining, in total agreement with the experiments. The radical product readily undergoes electron transfer to the metal center, which suggests that this intermediate is very unstable and thus very difficult to detect. The energy barrier computed for H atom transfer is 11.1 kcal mol-1 at the DFT(B3LYP) level and 13.6 kcal mol-1 at the IMOMM(B3LYP/MM3) level. These low energy barriers give further support to the reaction mechanism postulated for this enzymatic system. The complexes [M(L)] (M ) Zn(II), Cu(II); L ) N,N′bis(3,5-di-tert-butyl-2-hydroxyphenyl)-1,2-diiminoquinone) are functional models of galactose oxidase. The oxidation of primary alcohols with O2 catalyzed by these two complexes was studied by Wu at the DFT(B3LYP) level, including solvent effects.560 The quinone acts as a tetradentate

CsH Bond Activation in Transition Metal Species

ONNO chelating ligand and its oxygen atoms have radical phenoxyl character. In the first step of the mechanism, one of these oxygen atoms abstracts a proton from the alcohol substrate by breaking the OsH bond. In the following step, the other oxygen abstracts a radical hydrogen from the substrate by homolytic CsH cleavage. These two reactions weaken the metal-oxygen bonds involving the ligand. The calculations revealed a more favorable alternative pathway for the initial protonation step, in which an external base, triethylamine, takes the proton from the alcohol. H abstraction by the phenoxyl radical is the rate-determining step. The energy barriers computed for this step, 16.7 kcal mol-1 for Cu and 23.0 kcal mol-1 for Zn, are in good agreement with the experimental values, 19.5 and 21.1 kcal mol-1, respectively, and confirmed that the Zn catalyst is less efficient than the Cu one.

6. Conclusions Computational studies have considerably improved the perspective that chemists have on CsH bond activation by transition metal complexes. The past few years have seen significant advances in the field. In the past, where the theoretical methodologies and the computer speed imposed severe limitations on the size of the systems to calculate, computations were carried out on species that were heavily simplified relative to the real systems. Despite these limitations, computational studies had been very successful to quantify the role played by the metal centers on the barrier heights and energy of reactions. They also accounted successfully for the way electron-donating and electronwithdrawing ligands modify the electronic properties of the metal. As a result, calculations provided good insights in the fundamental aspects of the oxidative addition, σ bond metathesis and 1,2 addition. This resulted in an increasing role of computational studies devoted to reactivity and in particular to CsH activation as evidenced by the key reviews published in Chemical ReViews at the dawn of the 21st century.23–25 The ability to compute systems with a large number of atoms and thus species closer to the real systems did much more than just improve further the understanding of the well-established mechanisms. New mechanisms were discovered. For instance, σ-bond metathesis, which was originally considered almost exclusively for early transition metal complexes, was found to be also possible for late transition metal complexes, with the appropriate choice of ligands. However, only late metal complexes can form stable σ(CsH) bond complexes. To highlight the differences between early and late transition metal systems, the name σ-CAM was coined and computational studies have contributed to revealing the situations where this mechanism applies. The ability to calculate systems with large number of atoms has been of great importance to reveal the active role that ligands or various additives (base, counteranion) play in the CsH bond activation by assisting the CsH cleavage usually by way of a proton transfer. These mechanisms have been named CMD and AMLA. The calculations have shown that the assistance for the CsH activation can be achieved by a wide variety of ligands and that heterolytic cleavage of CsH is frequent. It is interesting to note that these new mechanisms, σ-CAM, CMD, and AMLA, lead to CsH activation occurring without change of oxidation state at the metal center and with low or moderate energy barriers. This is clearly an important direction to explore further for discovering new efficient ways to activate the CsH bond.

Chemical Reviews, 2010, Vol. 110, No. 2 817

The computational studies of CsH oxidation are also relatively recent because experimental systems engaging in CsH oxidation are often large, charged, or complex. The metal has often open-shell configurations, and several electronic states of the metal species can participate in the reaction. In a number of cases, the reactive species contain more than one metal, which makes the computational studies even more challenging. Even though DFT calculations of open shell systems need to be carried out with caution, there is no doubt that these calculations were able to suggest mechanisms and rationalize a considerable amount of experimental information. We refer the reader to the accompanying paper by Shaik, Thiel, et al. in this issue for the specific case of cytochrome P450 chemistry.50 It is rewarding that CsH activation is now the step of efficient catalytic processes. Computations have been and will be crucial to determine the rate-determining step in further catalytic processes. This information is required for improving the catalytic efficiency. The computational modeling of a reaction necessitates representing in an appropriate manner the experimental conditions (solvent, temperature, pressure). For instance, the modeling of the pressure of H2 gas was crucial to understand the acceptorless catalytic dehydrogenation of alkanes. The representation of experimental conditions is still a challenge for computational chemists but is required for future studies and especially for design of catalytic reactions. This review did not address the influence of the level of calculation on the energy profiles and mechanistic issues presented here because most of the studies have been carried out with a single method. Most of the authors have used a DFT method, which has proven its validity and solidity. Furthermore, most of the authors have used traditional functionals, which do not perform well to represent weak interactions. In a paradoxical manner, this may lead to a decrease of the quality of the calculations when considering complexes with the large real ligands because of the lack of representation of the weak interactions present in this case. This may not modify the essential features of the mechanism but may influence selectivity issues where interactions between ligands or between ligands and substrate could play a role. New functionals with the ability to represent weak interactions are being explored, and they will probably be more widely used in the future. Finally large systems could introduce conformational complexity, which may need also to be considered. In other words, it is very appealing to carry out computational studies of CsH bond activation with models similar to the experimental systems if the computational power is available, but the additional associated difficulties should not be forgotten. Finally, while there is still much more work to do to fully understand the selectivity of the CsH bond activation, it is clear that computational chemistry will be even more tightly associated with the experimental quest for new and more efficient transition metal catalysts.

7. Acknowledgments Our interest in CsH activation originates from many years of collaboration with experimental chemists. We would like to thank in particular Richard A. Andersen, Kenneth G. Caulton, Bruno Chaudret, Robert H. Crabtree, William D. Jones, Eduardo Peris, Robin N. Perutz, and Sylviane SaboEtienne for having shared their most exciting results with us and having educated us on the topic. Some of our own

818 Chemical Reviews, 2010, Vol. 110, No. 2

results are presented, and we are very grateful to Maria Besora, Marta Feliz, He´le`ne Ge´rard, Ulrike Helmstedt, Laurent Maron, Claire Me´gret, Lionel Perrin, and Christophe Raynaud for their contributions. D.B. thanks Sanofi-Aventis for a postdoctoral fellowship. O.E. thanks University of California, Berkeley, for its hospitality during the preparation of the review.

8. References (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14) (15) (16) (17) (18) (19) (20) (21) (22) (23) (24) (25) (26) (27) (28) (29)

(30) (31) (32) (33) (34) (35) (36) (37) (38) (39) (40) (41) (42) (43) (44) (45) (46)

Shilov, A. E.; Shul’pin, G. B. Chem. ReV. 1997, 97, 2879. Bergman, R. G. Science 1984, 223, 902. Jones, W. D.; Feher, F. J. Acc. Chem. Res. 1989, 22, 91. Crabtree, R. H. Chem. ReV. 1985, 85, 245. Crabtree, R. H. Chem. ReV. 1995, 95, 987. Arndtsen, B. A.; Bergman, R. G.; Mobley, T. A.; Peterson, T. H. Acc. Chem. Res. 1995, 28, 154. Jones, W. D. Top. Organomet. Chem. 1999, 3, 9. Crabtree, R. H. J. Chem. Soc., Dalton Trans. 2001, 2437. Labinger, J. A.; Bercaw, J. E. Nature 2002, 417, 507. Labinger, J. A. J. Mol. Catal. A 2004, 220, 27. Etienne, M.; McGrady, J. E.; Maseras, F. Coord. Chem. ReV. 2009, 253, 635. Dyker, G. Angew. Chem., Int. Ed. 1999, 38, 1698. Sen, A. Acc. Chem. Res. 1998, 31, 550. Guari, Y.; Sabo-Etienne, S.; Chaudret, B. Eur. J. Inorg. Chem. 1999, 1047. Jia, C.; Kitamura, T.; Fujiwara, Y. Acc. Chem. Res. 2001, 34, 633. Kakiuchi, F.; Murai, S. Acc. Chem. Res. 2002, 35, 826. Ritleng, V.; Sirlin, C.; Pfeffer, M. Chem. ReV. 2002, 102, 1731. Kakiuchi, F.; Chatani, N. AdV. Synth. Catal. 2003, 345, 1077. Jensen, C. M. Chem. Commun. 1999, 2443. Lewis, J. C.; Bergman, R. G.; Ellman, J. A. Acc. Chem. Res. 2008, 41, 1013. Godula, K.; Sames, D. Science 2006, 312, 67. Dedieu, A. Top. Organomet. Chem. 1999, 4, 69. Niu, S.; Hall, M. B. Chem. ReV. 2000, 100, 353. Torrent, M.; Sola`, M.; Frenking, G. Chem. ReV. 2000, 100, 439. Dedieu, A. Chem. ReV. 2000, 100, 543. Sakaki, S. Top. Organomet. Chem. 2005, 12, 31. Siegbahn, P. E. M. AdV. Chem. Phys. 1996, 93, 333. Go¨rling, A.; Trickey, S. B.; Gisdakis, P.; Ro¨sch, N. Top. Organomet. Chem. 1999, 4, 109. Authors report their results in terms of various energies: electronic energy (E), electronic energy corrected with zero-point energy (E + ZPE), enthalpy (H), and Gibbs free energy (G). Energy differences between a transition state and its associated reactant are referred to as energy barriers for ∆Eq, ∆(E + ZPE)q, and ∆Hq and as activation barriers for ∆Gq. The term activation energy refers to experimental values. Blomberg, M. R. A.; Siegbahn, P. E. M.; Nagashima, U.; Wennerberg, J. J. Am. Chem. Soc. 1991, 113, 424. Blomberg, M. R. A.; Siegbahn, P. E. M.; Svensson, M. J. Am. Chem. Soc. 1992, 114, 6095. Sautet, P.; Delbecq, F. Chem. ReV. 2010, 110, doi: 10.1021/ cr900295b. The electrophilic substitution mechanism is described in the section about Shilov chemistry. Groves, J. T. J. Chem. Educ. 1985, 62, 928. Murai, S.; Kakiuchi, F.; Sekine, S.; Tanaka, Y.; Kamatani, A.; Sonoda, M.; Chatani, N. Nature 1993, 366, 529. Das, S.; Brudvig, G. W.; Crabtree, R. H. Chem. Commun. 2008, 413. Gorelsky, S. I.; Lapointe, D.; Fagnou, K. J. Am. Chem. Soc. 2008, 130, 10848. Boutadla, Y.; Davies, D. L.; Macgregor, S. A.; Poblador-Bahamonde, A. L. Dalton Trans. 2009, 5820. Crabtree, R. H.; Mellea, M. F.; Mihelcic, J. M.; Quirk, J. M. J. Am. Chem. Soc. 1982, 104, 107. Xu, W. W.; Rosini, G. P.; Gupta, M.; Jensen, C. M.; Kaska, W. C.; Krogh-Jespersen, K.; Goldman, A. S. Chem. Commun. 1997, 2273. Liu, F. C.; Pak, E. B.; Singh, B.; Jensen, C. M.; Goldman, A. S. J. Am. Chem. Soc. 1999, 121, 4086. Goldman, A. S.; Roy, A. H.; Huang, Z.; Ahuja, R.; Schinski, W.; Brookhart, M. Science 2006, 312, 257. Periana, R. A.; Liu, X. Y.; Bhalla, G. Chem. Commun. 2002, 3000. Lail, M.; Arrowood, B. N.; Gunnoe, T. B. J. Am. Chem. Soc. 2003, 125, 7506. Waltz, K. M.; Hartwig, J. F. Science 1997, 277, 211. Chen, H. Y.; Schlecht, S.; Semple, T. C.; Hartwig, J. F. Science 2000, 287, 1995.

Balcells et al. (47) Goldshle, N. F.; Shteinmaa, A. A.; Shilov, A. E.; Eskova, V. V. Zh. Fiz. Khim. 1972, 46, 1353. (48) Periana, R. A.; Taube, D. J.; Evitt, E. R.; Loffler, D. G.; Wentrcek, P. R.; Voss, G.; Masuda, T. Science 1993, 259, 340. (49) Periana, R. A.; Taube, D. J.; Gamble, S.; Taube, H.; Satoh, T.; Fujii, H. Science 1998, 280, 560. (50) Shaik, S.; Cohen, S.; Wang, Y.; Chen, H.; Kumar, D.; Thiel, W. Chem. ReV. 2010, 110, doi: 10.1021/cr900121s. (51) Ortiz de Montellano, P. R. Chem. ReV. 2010, 110, doi: 10.1021/ cr9002193. (52) Wasserman, E. P.; Moore, C. B.; Bergman, R. G. Science 1992, 255, 315. (53) Hall, C.; Perutz, R. N. Chem. ReV. 1996, 96, 3125. (54) Geftakis, S.; Ball, G. E. J. Am. Chem. Soc. 1998, 120, 9953. (55) Cowan, A. J.; Portius, P.; Kawanami, H. K.; Jina, O. S.; Grills, D. C.; Sun, X.-Z.; McMaster, J.; George, M. W. Proc. Natl. Acad. Sci. U.S.A. 2007, 104, 6933. (56) Bernskoetter, W. H.; Schauer, C. K.; Goldberg, K. I.; Brookhart, M. Science 2009, 326, 553. (57) Lersch, M.; Tilset, M. Chem. ReV. 2005, 105, 2471. (58) Ziegler, T.; Tschinke, V.; Fan, V.-Y.; Becke, A. D. J. Am. Chem. Soc. 1989, 111, 9177. (59) Song, J.; Hall, M. B. Organometallics 1993, 12, 3118. (60) Musaev, D. G.; Morokuma, K. J. Am. Chem. Soc. 1995, 117, 799. (61) Siegbahn, P. E. M. J. Am. Chem. Soc. 1996, 118, 1487. (62) Su, M.-D.; Chu, S.-Y. Organometallics 1997, 16, 1621. (63) Jime´nez-Catan˜o, R.; Hall, M. B. Organometallics 1996, 15, 1889. (64) Su, M.-D.; Chu, S.-Y. J. Phys. Chem. A 1997, 101, 6798. (65) Su, M.-D.; Chu, S.-Y. Int. J. Quantum Chem. 1999, 72, 405. (66) Xavier, E. S.; De Alemeida, W. B.; da Silva, J. C. S.; Rocha, W. R. Organometallics 2005, 24, 2262. (67) Su, M.-D.; Chu, S.-Y. Chem.sEur. J. 1999, 5, 198. (68) Bi, S.; Zhang, Z.; Zhu, S. Chem. Phys. Lett. 2006, 431, 385. (69) Selmeczy, A. D.; Jones, W. D.; Partridge, M. G.; Perutz, R. N. Organometallics 1994, 12, 522. (70) Bosque, R.; Clot, E.; Fantacci, S.; Maseras, F.; Eisenstein, O.; Perutz, R. N.; Renkema, K. B.; Caulton, K. G. J. Am. Chem. Soc. 1998, 120, 12634. (71) Ge´rard, H.; Davidson, E. R.; Eisenstein, O. Mol. Phys. 2002, 100, 533. (72) Reinhold, M.; McGrady, J. E.; Perutz, R. N. J. Am. Chem. Soc. 2004, 126, 5268. (73) Nova, A.; Erhardt, S.; Jasim, N. A.; Perutz, R. N.; Macgregor, S. A.; McGrady, J. E.; Whitwood, A. C. J. Am. Chem. Soc. 2008, 130, 15499. (74) Erhardt, S.; Macgregor, S. A. J. Am. Chem. Soc. 2008, 130, 15490. (75) Stoutland, P. O.; Bergman, R. G. J. Am. Chem. Soc. 1985, 107, 4581. (76) Smith, K. M.; Poli, R.; Harvey, J. N. Chem.sEur. J. 2001, 7, 1679. (77) Harvey, J. N.; Aschi, M.; Schwarz, H.; Koch, W. Theor. Chem. Acc. 1998, 99, 95. (78) Harvey, J. N.; Aschi, M. Phys. Chem. Chem. Phys. 1999, 1, 5555. (79) Petit, A.; Richard, P.; Cacelli, I.; Poli, R. Chem.sEur. J. 2006, 12, 813. (80) Asplund, M. C.; Snee, P. T.; Yeston, J. S.; Wilkens, M. J.; Payne, C. K.; Yang, H.; Kotz, K. T.; Frei, H.; Bergman, R. G.; Harris, C. B. J. Am. Chem. Soc. 2002, 124, 10605. (81) Asbury, J. B.; Ghosh, H. N.; Yeston, J. S.; Bergman, R. G.; Lian, T. Organometallics 1998, 17, 3417. (82) Bromberg, S. E.; Yang, H.; Asplund, M. C.; Lian, T.; McNamara, B. K.; Kotz, K. T.; Yeston, J. S.; Wilkens, M.; Frei, H.; Bergman, R. G.; Harris, C. B. Science 1997, 278, 260. (83) Zaric´, S.; Hall, M. B. J. Phys. Chem. A 1998, 102, 1963. (84) Blake, A. J.; Georges, M. W.; Hall, M. B.; McMaster, J.; Portius, P.; Sun, X. Z.; Towrie, M.; Webster, C. E.; Wilson, C.; Zaric´, S. Organometallics 2008, 27, 189. (85) Clot, E.; Eisenstein, O.; Jones, W. D. Proc. Natl. Acad. Sci. U.S.A. 2007, 104, 6939. (86) Northcutt, T. O.; Wick, D. D.; Vetter, A. J.; Jones, W. D. J. Am. Chem. Soc. 2001, 123, 7257. (87) Jones, W. D. Acc. Chem. Res. 2003, 36, 140. (88) Ge´rard, H.; Eisenstein, O.; Lee, D.-H.; Chen, J.; Crabtree, R. H. New J. Chem. 2001, 25, 1121. (89) Iron, M. A.; Lo, H. C.; Martin, J. M. L.; Keinan, E. J. Am. Chem. Soc. 2002, 124, 7041. (90) Jensen, M. P.; Wick, D. D.; Reinartz, S.; White, P. S.; Templeton, J. L.; Goldberg, K. I. J. Am. Chem. Soc. 2003, 125, 8614. (91) Vastine, B. A.; Webster, C. E.; Hall, M. B. J. Chem. Theory Comput. 2007, 3, 2268. (92) Holtcamp, M. W.; Labinger, J. A.; Bercaw, J. E. J. Am. Chem. Soc. 1997, 119, 848. (93) Heiberg, H.; Johansson, L.; Gropen, O.; Ryan, O. B.; Swang, O.; Tilset, M. J. Am. Chem. Soc. 2000, 122, 10831.

CsH Bond Activation in Transition Metal Species (94) Johansson, L.; Tilset, M.; Labinger, J. A.; Bercaw, J. E. J. Am. Chem. Soc. 2000, 122, 10846. (95) Heiberg, H.; Swang, O.; Ryan, O. B.; Gropen, O. J. Phys. Chem. A 1999, 103, 10004. (96) Heiberg, H.; Gropen, O.; Swang, O. Int. J. Quantum Chem. 2003, 92, 391. (97) Li, J.-L.; Geng, C.-Y.; Huang, X.-R.; Zhang, X.; Sun, C.-C. Organometallics 2007, 26, 2203. (98) Nu¨ckel, S.; Bu¨rger, P. Angew. Chem., Int. Ed. 2003, 42, 1632. (99) Burger, P.; Bergman, R. G. J. Am. Chem. Soc. 1993, 115, 10462. (100) Tellers, D. M.; Yung, C. M.; Arndtsen, B. A.; Adamson, D. R.; Bergman, R. G. J. Am. Chem. Soc. 2002, 124, 1400. (101) Hinderling, C.; Feichtinger, D.; Plattner, D. A.; Chen, P. J. Am. Chem. Soc. 1997, 119, 10793. (102) Luecke, H. F.; Bergman, R. G. J. Am. Chem. Soc. 1997, 119, 11538. (103) Strout, D. L.; Zaric´, S.; Niu, S.; Hall, M. B. J. Am. Chem. Soc. 1996, 118, 6068. (104) Niu, S.; Hall, M. B. J. Am. Chem. Soc. 1998, 120, 6169. (105) Webster, C. E.; Hall, M. B. Organometallics 2001, 20, 5606. (106) Webster, C. E.; Hall, M. B. J. Organomet. Chem. 2003, 238-239, 315. (107) Viciano, M.; Feliz, M.; Corbera´n, R.; Mata, J. A.; Clot, E.; Peris, E. Organometallics 2007, 26, 5304. (108) Bergman, R. G.; Cundari, T. R.; Gillespie, A. M.; Gunnoe, T. B.; Harman, W. D.; Klinckman, T. R.; Temple, M. D.; White, D. P. Organometallics 2003, 22, 2331. (109) Clot, E.; Oelckers, B.; Klahn, A. H.; Eisenstein, O.; Perutz, R. N. Dalton Trans. 2003, 4065. (110) Carbo´, J. J.; Eisenstein, O.; Higgit, C. L.; Klahn, A. H.; Maseras, F.; Oelckers, B.; Perutz, R. N. Dalton Trans. 2001, 1452. (111) Lee, K.; Legzdins, P.; Pamplin, C. B.; Patrick, B. O.; Wada, K. Organometallics 2005, 24, 638. (112) Grebnik, P.; Downs, A. J.; Green, M. L. H.; Perutz, R. N. J. Chem. Soc., Chem. Commun. 1979, 742. (113) Chetwynd-Talbot, J.; Grebnik, P.; Perutz, R. N. Inorg. Chem. 1982, 21, 3647. (114) Gianotti, C.; Green, M. L. H. J. Chem. Soc., Chem. Commun. 1972, 1114. (115) Cooper, N. J.; Green, M. L. H.; Mahtab, R. J. Chem. Soc., Dalton Trans. 1979, 1557. (116) Berry, M.; Elmitt, K.; Green, M. L. H. J. Chem. Soc., Dalton Trans. 1979, 1950. (117) Bullock, R. M.; Headford, C. E. L.; Kegley, S. E.; Norton, J. R. J. Am. Chem. Soc. 1985, 107, 727. (118) Bullock, R. M.; Headford, C. E. L.; Hennessy, K. M.; Kegley, S. E.; Norton, J. R. J. Am. Chem. Soc. 1989, 111, 3897. (119) Parkin, G.; Bercaw, J. E. Organometallics 1989, 8, 1172. (120) Labella, L.; Chernega, A.; Green, M. L. H. J. Chem. Soc., Dalton Trans. 1995, 395. (121) Green, J. C.; Jardine, C. N. J. Chem. Soc., Dalton Trans. 1998, 1057. (122) Su, M.-D.; Chu, S.-Y. J. Phys. Chem. A 2001, 105, 3591. (123) Green, J. C.; Harvey, J. N.; Poli, R. J. Chem. Soc., Dalton Trans. 2002, 1861. (124) Green, J. C. Polyhedron 2005, 24, 1382. (125) Carreo´n-Macedo, J.-L.; Harvey, J. N.; Poli, R. Eur. J. Inorg. Chem. 2005, 2999. (126) Cox, P. A.; Grebenik, P.; Perutz, R. N.; Robinson, M. D.; Grinter, R.; Stern, D. Inorg. Chem. 1983, 22, 3614. (127) Grebenik, P.; Grinter, R.; Perutz, R. N. Chem. Soc. ReV. 1988, 453. (128) Shaik, S. J. Am. Chem. Soc. 1981, 103, 3692. (129) Churchill, D. G.; Janak, K. E.; Wittenberg, J. S.; Parkin, G. J. Am. Chem. Soc. 2003, 125, 1403. (130) Janak, K. E.; Parkin, G. J. Am. Chem. Soc. 2003, 125, 6889. (131) Parkin, G. Acc. Chem. Res. 2009, 42, 315. (132) Lin, Z. Coord. Chem. ReV. 2007, 251, 2280. (133) Ephritikhine, M. Chem. ReV. 1997, 97, 2193. (134) Kaminsky, W.; Arndt, M. AdV. Polym. Sci. 1997, 127, 143. (135) Burger, B. J.; Thompson, M. E.; Cotter, W. D.; Bercaw, J. E. J. Am. Chem. Soc. 1990, 112, 1566. (136) Thompson, M. E.; Buxter, S. M.; Bulls, A. R.; Burger, B. J.; Nolan, M. C.; Santarsiero, B. D.; Schaefer, W. P.; Bercaw, J. E. J. Am. Chem. Soc. 1987, 109, 203. (137) Margl, P.; Deng, L.; Ziegler, T. J. Am. Chem. Soc. 1999, 121, 154. (138) Watson, P. L.; Parshall, G. W. Acc. Chem. Res. 1985, 18, 51. (139) Davies, J. A.; Watson, P. L.; Liebman, J. F.; Greenberg, A. SelectiVe Hydrocarbon ActiVation:Principles and Progress; VCH Publishers: New York, 1990. (140) Werkema, E. L.; Messines, E.; Perrin, L.; Maron, L.; Eisenstein, O.; Andersen, R. A. J. Am. Chem. Soc. 2005, 127, 7781. (141) Werkema, E. L.; Andersen, R. A.; Yahia, A.; Maron, L.; Eisenstein, O. Organometallics 2009, 28, 3173. (142) Maron, L.; Werkema, E. L.; Perrin, L.; Eisenstein, O.; Andersen, R. A. J. Am. Chem. Soc. 2005, 127, 279.

Chemical Reviews, 2010, Vol. 110, No. 2 819 (143) Werkema, E. L.; Andersen, R. A. J. Am. Chem. Soc. 2008, 130, 7153. (144) Ma, K.; Piers, W. E.; Parvez, M. J. J. Am. Chem. Soc. 2006, 128, 3303. (145) Yang, X.; Hall, M. B. J. Phys. Chem. A 2009, 113, 2152. (146) Sadow, A. D.; Tilley, T. D. J. Am. Chem. Soc. 2003, 125, 7971. (147) Fontaine, F. G.; Tilley, T. D. Organometallics 2005, 24, 4340. (148) Steigerwald, M. L.; Goddard III, W. A. J. Am. Chem. Soc. 1984, 106, 308. (149) Rappe´, A. K. Organometallics 1990, 9, 466. (150) Ziegler, T.; Folga, E.; Berces, A. J. Am. Chem. Soc. 1993, 155, 636. (151) Maron, L.; Eisenstein, O. J. Am. Chem. Soc. 2001, 123, 1036. (152) Maron, L.; Eisenstein, O.; Alary, F.; Poteau, R. J. Phys. Chem. A 2002, 106, 1797. (153) Perrin, L.; Maron, L.; Eisenstein, O. New J. Chem. 2004, 28, 1255. (154) Maron, L.; Eisenstein, O. J. Phys. Chem. A 2000, 104, 7140. (155) Perrin, L.; Maron, L.; Eisenstein, O. ActiVation and Functionalization of C-H Bonds; Goldberg, K. I.; Goldman, A. S., Eds.; ACS Symposium Series 885; American Chemical Society: Washington, DC, 2004; p 116. (156) Folga, E.; Ziegler, T. Can. J. Chem. 1992, 70, 333. (157) Maron, L.; Perrin, L.; Eisenstein, O. J. Chem. Soc., Dalton Trans. 2002, 534. (158) Barros, N.; Eisentein, O.; Maron, L. Dalton Trans. 2006, 3052. (159) Ziegler, T.; Folga, E. J. Organomet. Chem. 1994, 478, 57. (160) Perrin, L.; Maron, L.; Eisenstein, O. Inorg. Chem. 2002, 41, 4355. (161) Perrin, L.; Eisenstein, O.; Maron, L. New J. Chem. 2007, 31, 549. (162) Perrin, L.; Maron, L.; Eisenstein, O.; Tilley, T. D. Organometallics 2009, 28, 3767. (163) Cope´ret, C.; Grouiller, A.; Basset, J.-M.; Chermette, H. ChemPhysChem 2003, 4, 608. (164) Sherer, E. C.; Cramer, C. J. Organometallics 2003, 22, 1682. (165) Woodrum, N. L.; Cramer, C. J. Organometallics 2006, 25, 68. (166) Lewin, J. L.; Woodrum, N. L.; Cramer, C. J. Organometallics 2006, 25, 5906. (167) Barros, N.; Eisenstein, O.; Maron, L.; Tilley, T. D. Organometallics 2006, 25, 5699. (168) Barros, N.; Eisenstein, O.; Maron, L.; Tilley, T. D. Organometallics 2008, 27, 2252. (169) Following this publication, the structure proposed by Hall could not be identified as a transition state. Maron, L. Private communication. (170) Perrin, L.; Maron, L.; Eisenstein, O. Dalton Trans. 2003, 4313. (171) Werkema, E. L.; Maron, L.; Eisenstein, O.; Andersen, R. A. J. Am. Chem. Soc. 2007, 129, 2529. (172) Guerrero-Rios, I.; Novarino, E.; van der Veer, S.; Hessen, B.; Bouwkamp, M. W. J. Am. Chem. Soc. 2009, 131, 16658. (173) Bennett, J. L.; Wolczanski, P. T. J. Am. Chem. Soc. 1997, 119, 10696. (174) Slaughter, L. M.; Wolczanski, P. T.; Klinckman, T. R.; Cundari, T. R. J. Am. Chem. Soc. 2000, 122, 7953. (175) Cundari, T. R.; Klinckman, T. R.; Wolczanski, P. T. J. Am. Chem. Soc. 2002, 124, 1481. (176) Ochi, N.; Nakao, Y.; Sato, H.; Sakaki, S. J. Am. Chem. Soc. 2007, 129, 8615. (177) Nakai, H.; Suzuki, J.; Kikuchi, Y. Chem. Phys. Lett. 2008, 460, 347. (178) Bailey, B. C.; Fan, H.; Baum, E. W.; Huffman, J. C.; Baik, M.-H.; Mindiola, D. J. J. Am. Chem. Soc. 2005, 127, 16016. (179) Bailey, B. C.; Fan, H.; Huffman, J. C.; Baik, M.-H.; Mindiola, D. J. J. Am. Chem. Soc. 2007, 129, 8781. (180) He, G.; Zhu, J.; Hung, W. Y.; Wen, T. B.; Sung, H. H.-Y.; Williams, I. D.; Lin, Z.; Jia, G. Angew. Chem., Int. Ed. 2007, 46, 9065. (181) Pamplin, C. B.; Legzdins, P. Acc. Chem. Res. 2003, 36, 223. (182) Poli, R.; Smith, K. M. Organometallics 2000, 19, 2858. (183) Adams, C. S.; Legzdins, P.; McNeil, W. S. Organometallics 2001, 20, 4939. (184) Fan, Y.; Hall, M. B. J. Chem. Soc., Dalton Trans. 2002, 713. (185) Wada, K.; Pamplin, C. B.; Legzdins, P. J. Am. Chem. Soc. 2002, 124, 9680. (186) Wada, K.; Pamplin, C. B.; Legzdins, P.; Patrick, B. O.; Tsyba, I.; Bau, R. J. Am. Chem. Soc. 2003, 125, 7035. (187) Fulton, J. R.; Holland, A. W.; Fox, D. J.; Bergman, R. G. Acc. Chem. Res. 2002, 35, 44. (188) Conner, D.; Jayaprakash, K. N.; Cundari, T. R.; Gunnoe, T. B. Organometallics 2004, 23, 2724. (189) Feng, Y.; Lail, M.; Barakat, K. A.; Cundari, T. R.; Gunnoe, T. B.; Petersen, J. L. J. Am. Chem. Soc. 2005, 127, 14174. (190) Feng, Y.; Gunnoe, T. B.; Grimes, T. V.; Cundari, T. R. Organometallics 2006, 25, 5456. (191) Feng, Y.; Lail, M.; Foley, N. A.; Gunnoe, T. B.; Barakat, K. A.; Cundari, T. R.; Petersen, J. L. J. Am. Chem. Soc. 2006, 128, 7982. (192) Cundari, T. R.; Grimes, T. V.; Gunnoe, T. B. J. Am. Chem. Soc. 2007, 129, 13172. (193) Tenn, W. J., III; Young, K. J. H.; Bhalla, G.; Oxgaard, J.; Goddard, W. A., III; Periana, R. A. J. Am. Chem. Soc. 2005, 127, 14172.

820 Chemical Reviews, 2010, Vol. 110, No. 2 (194) Tenn, W. J., III; Young, K. J. H.; Oxgaard, J.; Nielsen, R. J.; Goddard, W. A., III; Periana, R. A. Organometallics 2006, 25, 5173. (195) Oxgaard, J.; Tenn, W. J., III; Nielsen, R. J.; Periana, R. A.; Goddard, W. A., III Organometallics 2007, 26, 1565. (196) Perutz, R. N.; Sabo-Etienne, S. Angew. Chem., Int. Ed. 2007, 46, 2578. (197) Toner, A. J.; Gru¨ndemann, S.; Clot, E.; Limbach, H.-H.; Donnadieu, B.; Sabo-Etienne, S.; Chaudret, B. J. Am. Chem. Soc. 2000, 122, 6777. (198) Ng, S. M.; Lam, W. H.; Mak, C. C.; Tsang, C. W.; Jia, G.; Lin, Z.; Lau, C. P. Organometallics 2003, 22, 641. (199) Lam, W. H.; Jia, G.; Lin, Z.; Lau, C. P.; Eisenstein, O. Chem.sEur. J. 2003, 9, 2775. (200) Oxgaard, J.; Muller, R. P.; Goddard, W. A., III; Periana, R. A. J. Am. Chem. Soc. 2004, 126, 352. (201) Oxgaard, J.; Goddard, W. A., III J. Am. Chem. Soc. 2004, 126, 442. (202) DeYonker, N. J.; Foley, N. J.; Cundari, T. R.; Gunnoe, T. B.; Petersen, J. L. Organometallics 2007, 26, 6604. (203) Foley, N. A.; Gunnoe, T. B.; Cundari, T. R.; Boyle, P. D.; Petersen, J. L. Angew. Chem., Int. Ed. 2008, 47, 726. (204) Butschke, B.; Schro¨der, D.; Schwarz, H. Organometallics 2009, 28, 4340. (205) Lin, B.-L.; Bhattachryya, K. X.; Labinger, J. A.; Bercaw, J. E. Organometallics 2009, 28, 4400. (206) Bader, R. W. F. Atoms in Molecules, a Quantum Theory; Oxford University Press: Ithaca, NY, 1989. (207) Vastine, B. A.; Hall, M. B. J. Am. Chem. Soc. 2007, 129, 12068. (208) Vastine, B. A.; Hall, M. B. Coord. Chem. ReV. 2009, 253, 1202. (209) Fan, Y.; Hall, M. B. Organometallics 2005, 24, 3827. (210) Ess, D. H.; Nielsen, R. J.; Goddard, W. A., III; Periana, R. A. J. Am. Chem. Soc. 2009, 131, 11686. (211) Khaliullin, R. Z.; Cobar, E. A.; Lochan, R. C.; Bell, A. T.; HeadGordon, M. J. Phys. Chem. A 2007, 111, 8753. (212) Metal-Oxo and Metal-Peroxo Species in Catalytic Oxidations (Structure and Bonding); Meunier, B., Ed.; Springer-Verlag: Berlin, 2000. (213) Catalytic ActiVation and Functionalisation of Light Alkanes- AdVances and Challenges; Derouane, E. G., Haber, J., Lemos, F., Ribeiro, F. R., Guisnet, M., Eds.; Springer: Dordrecht, The Netherlands, 1998. (214) Biomimetic Oxidations Catalyzed by Transition Metal Complexes; Meunier, B., Ed.; Imperial College Press: London, 2000. (215) Modern Biooxidation: Enzymes, Reactions and Applications; Schmid, R. D., Urlacher, V., Eds.; Wiley-VCH: Weinheim, Germany, 2007. (216) Solomon, E. I.; Sundaram, U. M.; Machonkin, T. E. Chem. ReV. 1996, 96, 2563. (217) Que, L.; Dong, Y. H. Acc. Chem. Res. 1996, 29, 190. (218) Cytochrome P450: Structure, Mechanism, and Biochemistry; de Montellano, P. R. O., Ed.; Springer: New York, 2005. (219) Groves, J. T.; McClusky, G. A. J. Am. Chem. Soc. 1976, 98, 859. (220) Das, S.; Incarvito, C. D.; Crabtree, R. H.; Brudvig, G. W. Science 2006, 312, 1941. (221) Denisov, I. G.; Makris, T. M.; Sligar, S. G.; Schlichting, I. Chem. ReV. 2005, 105, 2253. (222) Loew, G. H.; Harris, D. L. Chem. ReV. 2000, 100, 407. (223) Dawson, J. H.; Sono, M. Chem. ReV. 1987, 87, 1255. (224) Meunier, B.; de Visser, S. P.; Shaik, S. Chem. ReV. 2004, 104, 3947. (225) Shaik, S.; Kumar, D.; de Visser, S. P.; Altun, A.; Thiel, W. Chem. ReV. 2005, 105, 2279. (226) Costas, M.; Mehn, M. P.; Jensen, M. P.; Que, L. Chem. ReV. 2004, 104, 939. (227) Que, L. Acc. Chem. Res. 2007, 40, 493. (228) Que, L.; Ho, R. Y. N. Chem. ReV. 1996, 96, 2607. (229) Carlsen, C. U.; Moller, J. K. S.; Skibsted, L. H. Coord. Chem. ReV. 2005, 249, 485. (230) Hammes-Schiffer, S. Acc. Chem. Res. 2006, 39, 93. (231) Fenton, H. J. H. J. Chem. Soc. Trans. 1894, 65, 899. (232) Koch, W.; Holthausen, M. C. A Chemist’s Guide to Density Functional Theory; Wiley-VCH: Weinheim, Germany, 2000. (233) Maseras, F.; Morokuma, K. J. Comput. Chem. 1995, 16, 1170. (234) Uddin, J.; Morales, C. M.; Maynard, J. H.; Landis, C. R. Organometallics 2006, 25, 5566. (235) Shilov, A. E. Alkane ActiVation Complexes of Rhodium, Iridium and Related Species; Wiley: New York, 1989; p 372. (236) Hong, S.; Marks, T. J. Acc. Chem. Res. 2004, 37, 673. (237) Martinho-Simo˜es, J. A.; Beauchamp, J. L. Chem. ReV. 1990, 90, 629. (238) Hoff, C. D. Prog. Inorg. Chem. 1992, 40, 503. (239) King, W. A.; DiBella, S.; Gulino, A.; Lanza, G.; Fragala´, I. L.; Stern, C. L.; Marks, T. J. J. Am. Chem. Soc. 1999, 121, 355. (240) Bryndza, H. E.; Fong, L. K.; Paciello, R. A.; Tam, W.; Bercaw, J. E. J. Am. Chem. Soc. 1987, 109, 1444. (241) Jones, W. D.; Hessell, E. T. J. Am. Chem. Soc. 1993, 115, 554. (242) Wick, D. D.; Jones, W. D. Organometallics 1999, 18, 495. (243) Bennett, J. L.; Wolczanski, P. T. J. Am. Chem. Soc. 1994, 116, 2179.

Balcells et al. (244) Holland, P. L.; Andersen, R. A.; Bergman, R. G.; Huang, J.; Nolan, S. P. J. Am. Chem. Soc. 1997, 119, 12800. (245) Holland, P. L.; Andersen, R. A.; Bergman, R. G. Comments Inorg. Chem. 1999, 21, 115. (246) Schock, L. E.; Marks, T. J. J. Am. Chem. Soc. 1988, 110, 7701. (247) Yao, X.-Q.; Hou, X.-J.; Jiao, H.; Xiang, H.-W.; Li, Y.-W. J. Phys. Chem. A 2003, 107, 9991. (248) Zhao, Y.; Lynch, B. J.; Truhlar, D. G. J. Phys. Chem. A 2004, 108, 2715. (249) Zhao, Y.; Lynch, B. J.; Truhlar, D. G. J. Phys. Chem. A 2004, 108, 4786. (250) Izgorodina, E. I.; Coote, M. L.; Radom, L. J. Phys. Chem. A 2005, 109, 7558. (251) Gillies, M. B.; Matyjaszewski, K.; Norrby, P. O.; Pintauer, T.; Poli, R.; Richard, P. Macromolecules 2003, 36, 8551. (252) Henry, D. J.; Parkinson, C. J.; Mayer, P. M.; Radom, L. J. Phys. Chem. A 2001, 105, 6750. (253) Zhao, Y.; Truhlar, D. G. Acc. Chem. Res. 2008, 41, 157. (254) Schwabe, T.; Grimme, S. Acc. Chem. Res. 2008, 41, 569. (255) Ziegler, T.; Tschinke, V.; Becke, A. J. Am. Chem. Soc. 1987, 109, 1351. (256) Folga, E.; Ziegler, T. J. Am. Chem. Soc. 1993, 115, 5169. (257) Ziegler, T.; Cheng, W.; Baerends, E. J.; Ravanek, W. Inorg. Chem. 1988, 27, 3458. (258) Harvey, J. N. Organometallics 2001, 20, 4887. (259) Clot, E.; Me´gret, C.; Eisenstein, O.; Perutz, R. N. J. Am. Chem. Soc. 2006, 128, 8350. (260) Mitoraj, M.; Zhu, H.; Michalak, A.; Ziegler, T. Organometallics 2007, 26, 1627. (261) Clot, E.; Besora, M.; Maseras, F.; Me´gret, C.; Eisenstein, O.; Oelckers, B.; Perutz, R. N. Chem. Commun. 2003, 490. (262) Clot, E.; Me´gret, C.; Eisenstein, O.; Perutz, R. N. J. Am. Chem. Soc. 2009, 131, 7817. (263) Evans, M. E.; Burke, C. L.; Yaibuathes, S.; Clot, E.; Eisenstein, O.; Jones, W. D. J. Am. Chem. Soc. 2009, 131, 13464. (264) Diggle, R. A.; Kennedy, A. A.; Macgregor, S. A.; Whittlesey, M. K. Organometallics 2008, 27, 938. (265) Matsubara, T.; Koga, N.; Musaev, D.; Morokuma, K. J. Am. Chem. Soc. 1998, 120, 12692. (266) Guari, Y.; Sabo-Etienne, S.; Chaudret, B. J. Am. Chem. Soc. 1998, 120, 4228. (267) Toner, A.; Matthes, J.; Gru¨ndemann, S.; Limbach, H.-H.; Chaudret, B.; Clot, E.; Sabo-Etienne, S. Proc. Natl. Acad. Sci. U.S.A. 2007, 104, 6945. (268) Barrio, P.; Castarlenas, R.; Esteruelas, M. A.; Lledo´s, A.; Maseras, F.; On˜ate, E.; Tomàs, J. Organometallics 2001, 20, 442. (269) Esteruelas, M. A.; Lledo´s, A.; Oliva´n, M.; On˜ate, E.; Tajada, M. A.; Ujaque, G. Organometallics 2003, 22, 3753. (270) Ben-Ari, E.; Gandelman, M.; Rozenberg, H.; Shimon, L. J. W.; Milstein, D. J. Am. Chem. Soc. 2003, 125, 4714. (271) Ben-Ari, E.; Cohen, R.; Gandelman, M.; Shimon, L. J. W.; Martin, J. M. L.; Milstein, D. Organometallics 2006, 25, 3190. (272) Fan, L.; Parkin, S.; Ozerov, O. V. J. Am. Chem. Soc. 2005, 127, 16772. (273) Wu, H.; Hall, M. B. Dalton Trans. 2009, 5933. (274) Wu, H.; Hall, M. B. J. Phys. Chem. A 2009, 113, 11706. (275) Zhang, X.; Kanzelberger, M.; Emge, T. J.; Goldman, A. S. J. Am. Chem. Soc. 2004, 126, 13192. (276) Zhang, X.; Emge, T. J.; Ghosh, R.; Krogh-Jespersen, K.; Goldman, A. S. Organometallics 2006, 25, 1303. (277) Feller, M.; Karton, A.; Leitus, G.; Martin, J. M. L.; Milstein, D. J. Am. Chem. Soc. 2006, 128, 12400. (278) Paneque, M.; Poveda, M. L.; Santos, L. L.; Carmona, E.; Lledo´s, A.; Ujaque, G.; Mereiter, K. Angew. Chem., Int. Ed. 2004, 43, 3708. (279) Lara, P.; Paneque, M.; Poveda, M. L.; Santos, L. L.; Valpuesta, ´ lvarez, J. E. V.; Carmona, E.; Moncho, S.; Ujaque, G.; Lledo´s, A.; A E.; Mereiter, K. Chem.sEur. J. 2009, 15, 9034. (280) Lara, P.; Paneque, M.; Poveda, M. L.; Santos, L. L.; Valpuesta, J. E. V.; Salazar, V.; Carmona, E.; Moncho, S.; Ujaque, G.; Lledo´s, A.; Maya, C.; Mereiter, K. Chem.sEur. J. 2009, 15, 9046. (281) Jordan, R. F.; Taylor, D. F. J. Am. Chem. Soc. 1989, 111, 778. (282) Bi, S.; Lin, Z.; Jordan, R. F. Organometallics 2004, 23, 4882. (283) Pool, J. A.; Scott, B. L.; Kiplinger, J. L. J. Alloys Compd. 2006, 418, 178. (284) Kiplinger, J. L.; Scott, B. L.; Schelter, E. J.; Pool, J. A. J. Alloys Compd. 2007, 444-445, 477. (285) Yang, P.; Warnke, I.; Martin, R. L.; Hay, P. J. Organometallics 2008, 27, 1384. (286) Tan, K. L.; Bergman, R. G.; Ellman, J. A. J. Am. Chem. Soc. 2002, 124, 3202. (287) Wiedemann, S. H.; Lewis, J. C.; Ellman, J. A.; Bergman, R. G. J. Am. Chem. Soc. 2006, 128, 2452.

CsH Bond Activation in Transition Metal Species (288) Yao, W. B.; Eisenstein, O.; Crabtree, R. H. Inorg. Chem. Acta 1997, 254, 105. (289) Baya, M.; Eguillor, B.; Esteruelas, M. A.; Lledo´s, A.; Oliva´n, M.; On˜ate, E. Organometallics 2007, 26, 5140. (290) Esteruelas, M. A.; Ferna´ndez-Alvarez, F. J.; On˜ate, E. J. Am. Chem. Soc. 2006, 128, 13044. (291) Esteruelas, M. A.; Force´n, E.; Oliva´n, M.; On˜ate, E. Organometallics 2008, 27, 6188. (292) Esteruelas, M. A.; Ferna´ndez-Alverez, F. J.; On˜ate, E. Organometallics 2008, 27, 6236. (293) Eguillor, B.; Esteruelas, M. A.; Garcı´a-Raboso, J.; Oliva´n, M.; On˜ate, E. Organometallics 2009, 28, 3700. (294) Clot, E.; Chen, J.; Lee, D.-H.; Sung, S. Y.; Appelhans, L. N.; Faller, J. W.; Crabtree, R. H.; Eisenstein, O. J. Am. Chem. Soc. 2004, 126, 8795. (295) van der Boom, M.; Milstein, D. Chem. ReV. 2003, 103, 1759. (296) Diez-Gonza´lez, S.; Marion, N.; Nolan, S. P. Chem. ReV. 2009, 109, 3612. (297) Vigalok, A.; Uzan, O.; Shimon, L. J. W.; Ben-David, Y.; Martin, J. M. L.; Milstein, D. J. Am. Chem. Soc. 1998, 120, 12539. (298) Montag, M.; Schwartsburd, L.; Cohen, R.; Leitus, G.; Ben-David, Y.; Martin, J. M. L.; Milstein, D. Angew. Chem., Int. Ed. 2007, 46, 1901. (299) Rybtchinski, B.; Cohen, R.; Ben-David, Y.; Martin, J. M. L.; Milstein, D. J. Am. Chem. Soc. 2003, 125, 11041. (300) Rybtchinski, B.; Konstantinovsky, L.; Shimon, L. J. W.; Vigalok, A.; Milstein, D. Chem.sEur. J. 2000, 6, 3287. (301) Blug, M.; Heuclin, H.; Cantat, T.; Le Goff, X.-F.; Me´zailles, N.; Le Floch, P. Organometallics 2009, 28, 1969. (302) Cantat, T.; Demange, M.; Me´zailles, N.; Ricard, L.; Jean, Y.; Le Floch, P. Organometallics 2005, 24, 4838. (303) Jazzar, R. F. R.; Macgregor, S. A.; Mahon, M. F.; Richards, S. P.; Whittlesey, M. K. J. Am. Chem. Soc. 2002, 124, 4944. (304) Diggle, R. A.; Macgregor, S. A.; Whittlesey, M. K. Organometallics 2008, 27, 617. (305) Mathew, J.; Koga, N.; Suresh, C. H. Organometallics 2008, 27, 4666. (306) Diggle, R. A.; Macgregor, S. A.; Whittlesey, M. K. Organometallics 2004, 23, 1857. (307) Poyatos, M.; Mata, J. A.; Peris, E. Chem. ReV. 2009, 109, 3677. (308) Viciano, M.; Poyatos, M.; Sanau´, M.; Peris, E.; Rossin, A.; Ujaque, G.; Lledo´s, A. Organometallics 2006, 25, 1120. (309) Kovacevic, A.; Gru¨ndemann, S.; Miecznikowski, J. R.; Clot, E.; Eisenstein, O.; Crabtree, R. H. Chem. Commun. 2002, 2580. (310) Appelhans, L. N.; Zuccaccia, D.; Kovacevic, A.; Chianese, A. R.; Miecznikowski, J. R.; Macchioni, A.; Clot, E.; Eisenstein, O.; Crabtree, R. H. J. Am. Chem. Soc. 2005, 127, 16299. (311) Clot, E. Eur. J. Inorg. Chem. 2009, 2319. (312) Ha¨ller, L. J. L.; Page, M. J.; Macgregor, S. A.; Mahon, M. F.; Whittlesey, M. K. J. Am. Chem. Soc. 2009, 131, 4604. (313) Marrone, A.; Re, N.; Romeo, R. Organometallics 2008, 27, 2215. (314) Kuznetsov, V. F.; Abdur-Rashid, K.; Lough, A. J.; Gusev, D. G. J. Am. Chem. Soc. 2006, 128, 14388. (315) Ingleson, M. J.; Yang, X.; Pink, M.; Caulton, K. G. J. Am. Chem. Soc. 2005, 127, 10846. (316) Walstrom, A.; Pink, M.; Tsvetkov, N. P.; Fan, H.; Ingleson, M. J.; Caulton, K. G. J. Am. Chem. Soc. 2005, 127, 16780. (317) Das, S.; Brudvig, G. W.; Crabtree, R. H. J. Am. Chem. Soc. 2008, 130, 1628. (318) Balcells, D.; Moles, P.; Blakemore, J.; Raynaud, C.; Brudvig, G. W.; Crabtree, R. H.; Eisenstein, O. Dalton Trans. 2009, 5989. (319) Lundberg, M.; Blomberg, M. R. A.; Siegbahn, P. E. M. Inorg. Chem. 2004, 43, 264. (320) Balcells, D.; Raynaud, C.; Crabtree, R. H.; Eisenstein, O. Chem. Commun. 2008, 744. (321) Balcells, D.; Raynaud, C.; Crabtree, R. H.; Eisenstein, O. Inorg. Chem. 2008, 47, 10090. (322) Balcells, D.; Raynaud, C.; Crabtree, R. H.; Eisenstein, O. Chem. Commun. 2009, 1772. (323) Biswas, B.; Sugimoto, M.; Sakaki, S. Organometallics 2000, 19, 3895. (324) Davies, D. L.; Donald, S. M. A.; Macgregor, S. A. J. Am. Chem. Soc. 2005, 127, 13754. (325) Davies, D. L.; Donald, S. M. A.; Al-Duaij, O.; Macgregor, S. A.; Po¨lleth, M. J. Am. Chem. Soc. 2006, 128, 4210. (326) Boutadla, Y.; Davies, D. L.; Macgregor, S. A.; Poblador-Bahamonde, A. I. Dalton Trans. 2009, 5887. (327) Davies, D. L.; Donald, S. M. A.; Al-Duaij, O.; Fawcett, J.; Little, C.; Macgregor, S. A. Organometallics 2006, 25, 5976. (328) Bielsa, R.; Navarro, R.; Urriolabeitia, E. P.; Lledo´s, A. Inorg. Chem. 2007, 46, 10133. (329) Aguilar, D.; Bielsa, R.; Contel, M.; Lledo´s, A.; Navarro, R.; Soler, T.; Urriolabeitia, E. P. Organometallics 2008, 27, 2929. (330) Aullo´n, G.; Chat, R.; Favier, I.; Font-Bardia, M.; Go´mez, M.; Granell, J.; Martı´nez, M.; Solans, X. Dalton Trans. 2009, 8292.

Chemical Reviews, 2010, Vol. 110, No. 2 821 (331) Ess, D. H.; Bischof, S. M.; Oxgaard, J.; Periana, R. A.; Goddard III, W. A. Organometallics 2008, 27, 6440. (332) Garcı´a-Cuadrado, D.; Braga, A. A. C.; Maseras, F.; Echavarren, A. M. J. Am. Chem. Soc. 2006, 128, 1066. (333) García-Cuadrado, D.; de Mendoza, P.; Braga, A. A. C.; Maseras, F.; Echavarren, A. M. J. Am. Chem. Soc. 2007, 129, 6880. (334) Pascual, S.; de Mendoza, P.; Braga, A. A. C.; Maseras, F.; Echavarren, A. M. Tetrahedron 2008, 64, 6021. (335) Campeau, L.-C.; Fagnou, K. Chem. Commun. 2006, 1253. (336) Lafrance, M.; Rowley, C. N.; Woo, T. K.; Fagnou, K. J. Am. Chem. Soc. 2006, 128, 8754. ¨ zdemir, I.; Demir, S.; C¸etinkaya, B.; Gourlaouen, C.; Maseras, F.; (337) O Bruneau, C.; Dixneuf, P. H. J. Am. Chem. Soc. 2008, 130, 1156. (338) Lafrance, M.; Gorelsky, S. I.; Fagnou, K. J. Am. Chem. Soc. 2007, 129, 14570. (339) Chaumontet, M.; Piccardi, R.; Audic, N.; Hitce, J.; Peglion, J.-L.; Clot, E.; Baudoin, O. J. Am. Chem. Soc. 2008, 130, 15157. (340) Baudry, D.; Ephritikine, M.; Felkin, H.; Holmes-Smith, R. J. Chem. Soc., Chem. Commun. 1983, 788. (341) Felkin, H.; Fillebeen-Khan, T.; Gault, Y.; Holmes-Smith, R.; Zakrzewski, J. Tetrahedron Lett. 1984, 25, 1279. (342) Burk, M. J.; Crabtree, R. H.; Parnell, C. P.; Uriarte, R. J. Organometallics 1984, 3, 816. (343) Burk, M. J.; Crabtree, R. H.; McGrath, D. V. J. Chem. Soc., Chem Commun. 1985, 1829. (344) Maguire, J. A.; Goldman, A. S. J. Am. Chem. Soc. 1991, 113, 6706. (345) Maguire, J. A.; Petrillo, A.; Goldman, A. S. J. Am. Chem. Soc. 1992, 114, 9492. (346) Gupta, M.; Hagen, C.; Flesher, R. J.; Kaska, W. C.; Jensen, C. M. Chem. Commun. 1996, 2083. (347) Gupta, M.; Hagen, C.; Kaska, W. C.; Cramer, R. E.; Jensen, C. M. J. Am. Chem. Soc. 1997, 119, 840. (348) Fujii, T.; Saito, Y. J. Chem. Soc., Chem. Commun. 1990, 757. (349) Aoki, T.; Crabtree, R. H. Organometallics 1993, 12, 294. (350) Liu, F.; Goldman, A. S. Chem. Commun. 1999, 655. (351) Li, S.; Hall, M. B. Organometallics 2001, 20, 3210. (352) Li, S.; Hall, M. B. Organometallics 2001, 20, 2153. (353) Krogh-Jespersen, K.; Czerw, M.; Kanzelberger, M.; Goldman, A. S. J. Chem. Inf. Comput. Sci. 2001, 41, 56. (354) Haenel, M. W.; Oevers, S.; Angermund, K.; Kaska, W. C.; Fan, H.J.; Hall, M. B. Angew. Chem., Int. Ed. 2001, 40, 3596. (355) Krogh-Jespersen, K.; Czerw, M.; Summa, N.; Renkema, K. B.; Achord, P. D.; Goldman, A. S. J. Am. Chem. Soc. 2002, 124, 11404. (356) Fan, H.-J.; Hall, M. B. J. Mol. Catal. A 2002, 189, 111. (357) Krogh-Jespersen, K.; Czerw, M.; Goldman, A. S. J. Mol. Catal. A 2002, 189, 95. (358) Zhu, K.; Achord, P. D.; Zhang, X.; Krogh-Jespersen, K.; Goldman, A. S. J. Am. Chem. Soc. 2004, 126, 13044. (359) Kundu, S.; Choliy, Y.; Zhuo, G.; Ahuja, R.; Emge, T. J.; Warmuth, R.; Brookhart, M.; Krogh-Jespersen, K.; Goldman, A. S. Organometallics 2009, 28, 5432. (360) Chaplin, A. B.; Poblador-Bahamonde, A. I.; Sparkes, H. A.; Howard, J. A. K.; Macgregor, S. A.; Weller, A. S. Chem. Commun. 2009, 244. (361) Douglas, T. M.; Brayshaw, S. K.; Dallanegra, R.; Kociok-Ko¨hn, G.; Macgregor, S. A.; Moxham, G. K.; Weller, A. S.; Wondimagegn, T.; Vadivelu, P. Chem.sEur. J. 2008, 14, 1004. (362) Douglas, T. M.; Weller, A. S. New J. Chem. 2008, 32, 966. (363) Grellier, M.; Vendier, L.; Sabo-Etienne, S. Angew. Chem., Int. Ed. 2007, 46, 2613. (364) Bolton, P. D.; Grellier, M.; Vautravers, N.; Vendier, L.; Sabo-Etienne, S. Organometallics 2008, 27, 5088. (365) Matsubara, T.; Koga, N.; Musaev, D. G.; Morokuma, K. Organometallics 2000, 19, 2318. (366) Matsumoto, T.; Taube, D. J.; Periana, R. A.; Taube, H.; Yoshida, H. J. Am. Chem. Soc. 2000, 122, 7414. (367) Matsumoto, T.; Periana, R. A.; Taube, D. J.; Yoshida, H. J. Mol. Catal. A 2002, 180, 1. (368) Bhalla, G.; Liu, X. Y.; Oxgaard, J.; Goddard, W. A., III; Periana, R. A. J. Am. Chem. Soc. 2005, 127, 11372. (369) Bhalla, G.; Oxgaard, J.; Goddard, W. A., III; Periana, R. A. Organometallics 2005, 24, 3229. (370) Bhalla, G.; Oxgaard, J.; Goddard, W. A., III; Periana, R. A. Organometallics 2005, 24, 5499. (371) Oxgaard, J.; Bhalla, G.; Periana, R. A.; Goddard, W. A., III Organometallics 2006, 25, 1618. (372) Oxgaard, J.; Periana, R. A.; Goddard, W. A., III J. Am. Chem. Soc. 2004, 126, 11658. (373) Lail, M.; Bell, C. M.; Conner, D.; Cundari, T. R.; Gunnoe, T. B.; Petersen, J. L. Organometallics 2004, 23, 5007. (374) Pittard, K. A.; Lee, J. P.; Cundari, T. R.; Gunnoe, T. B.; Petersen, J. L. Organometallics 2004, 23, 5514.

822 Chemical Reviews, 2010, Vol. 110, No. 2 (375) Pittard, K. A.; Cundari, T. R.; Gunnoe, T. B.; Day, C. S.; Petersen, J. L. Organometallics 2005, 24, 5015. (376) Foley, N. A.; Lail, M.; Lee, J. P.; Gunnoe, T. B.; Cundari, T. R.; Petersen, J. L. J. Am. Chem. Soc. 2007, 129, 6765. (377) Foley, N. A.; Lail, M.; Gunnoe, T. B.; Cundari, T. R.; Boyle, P. D.; Petersen, J. L. Organometallics 2007, 26, 5507. (378) Foley, N. A.; Ke, Z.; Gunnoe, T. B.; Cundari, T. R.; Petersen, J. L. Organometallics 2008, 27, 3007. (379) Foley, N. A.; Lee, J. P.; Ke, Z.; Gunnoe, T. B.; Cundari, T. R. Acc. Chem. Res. 2009, 42, 585. (380) Iverson, C. N.; Smith, M. R., III J. Am. Chem. Soc. 1999, 121, 7696. (381) Wan, X.; Wang, X.; Luo, Y.; Takami, S.; Kubo, M.; Miyamoto, A. Organometallics 2002, 21, 3703. (382) Webster, C. E.; Fan, Y.; Hall, M. B.; Kunz, D.; Hartwig, J. F. J. Am. Chem. Soc. 2003, 125, 858. (383) Hartwig, J. F.; Cook, K. S.; Hapke, M.; Incarvito, C. D.; Fan, Y.; Webster, C. E.; Hall, M. B. J. Am. Chem. Soc. 2005, 127, 2538. (384) Lam, W. H.; Lin, Z. Organometallics 2003, 22, 473. (385) Lam, W. H.; Lam, K. C.; Lin, Z.; Shimada, S.; Perutz, R. N.; Marder, T. B. Dalton Trans. 2004, 1556. (386) Dang, L.; Lin, Z.; Marder, T. B. Chem. Commun. 2009, 3987. (387) Tamura, H.; Yamazaki, H.; Sato, H.; Sakaki, S. J. Am. Chem. Soc. 2003, 125, 16114. (388) Ishiyama, T.; Takagi, J.; Ishida, K.; Miyaura, N.; Anastasi, N. R.; Hartwig, J. F. J. Am. Chem. Soc. 2002, 124, 390. (389) Ishiyama, T.; Takagi, J.; Hartwig, J. F.; Miyaura, N. Angew. Chem., Int. Ed. 2002, 41, 3056. (390) Chotana, G. A.; Vanchura, B. A.; Tse, M. T.; Staples, R. J.; Maleczka, R. E.; Smith, M. R., III Chem. Commun. 2009, 5731. (391) Garnett, J. L.; Hodges, R. J. J. Am. Chem. Soc. 1967, 89, 4546. (392) Shilov, A. E. ActiVation of Saturated Hydrocarbons by Transition Metal Complexes; Springer: Dordrecht, The Netherlands, 1984. (393) Siegbahn, P. E. M.; Crabtree, R. H. J. Am. Chem. Soc. 1996, 118, 4442. (394) Eisenstein, O.; Crabtree, R. H. New J. Chem. 2001, 25, 665. (395) Zhu, H.; Ziegler, T. J. Organomet. Chem. 2006, 691, 4486. (396) Zhu, H.; Ziegler, T. Organometallics 2007, 26, 2277. (397) Zhu, H.; Ziegler, T. Organometallics 2008, 27, 1743. (398) Zhu, H.; Ziegler, T. Organometallics 2009, 28, 2773. (399) Arakawa, H.; Aresta, M.; Armor, J. N.; Barteau, M. A.; Beckman, E. J.; Bell, A. T.; Bercaw, J. E.; Creutz, C.; Dinjus, E.; Dixon, D. A.; Domen, K.; DuBois, D. L.; Eckert, J.; Fujita, E.; Gibson, D. H.; Goddard, W. A., III; Goodman, D. W.; Keller, J.; Kubas, G. J.; Kung, H. H.; Lyons, J. E.; Manzer, L. E.; Marks, T. J.; Morokuma, K.; Nicholas, K. M.; Periana, R.; Que, L.; Rostrup-Nielson, J.; Sachtler, W. M. H.; Schmidt, L. D.; Sen, A.; Somorjai, G. A.; Stair, P. C.; Stults, B. R.; Tumas, W. Chem. ReV. 2001, 101, 953. (400) Periana, R. A.; Bhalla, G.; Tenn, W. J., III; Young, K. J. H.; Liu, X. Y.; Mironov, O.; Jones, C. J.; Ziatdinov, V. R. J. Mol. Catal. A 2004, 220, 7. (401) Conley, B. L.; Tenn, W. J., III; Young, K. J. H.; Ganesh, S. K.; Meier, S. K.; Ziatdinov, V. R.; Mironov, O. A.; Oxgaard, J.; Gonzales, J. M.; Goddard, W. A., III; Periana, R. A. J. Mol. Catal. A 2006, 251, 8. (402) Mylvaganam, K.; Bacskay, G. B.; Hush, N. S. J. Am. Chem. Soc. 1999, 121, 4633. (403) Mylvaganam, K.; Bacskay, G. B.; Hush, N. S. J. Am. Chem. Soc. 2000, 122, 2041. (404) Gilbert, T. M.; Hristov, J.; Ziegler, T. Organometallics 2001, 20, 1183. (405) Kua, J.; Xu, X.; Periana, R. A.; Goddard, W. A., III Organometallics 2002, 21, 511. (406) Hristov, I. H.; Ziegler, T. Organometallics 2003, 22, 1668. (407) Xu, X.; Kua, J.; Periana, R. A.; Goddard, W. A., III Organometallics 2003, 22, 2057. (408) Paul, A.; Musgrave, C. B. Organometallics 2007, 26, 793. (409) Ahlquist, M.; Periana, R. A.; Goddard, W. A., III Chem. Commun. 2009, 2373. (410) Ziatdinov, V. R.; Oxgaard, J.; Mironov, O. A.; Young, K. J. H.; Goddard, W. A., III; Periana, R. A. J. Am. Chem. Soc. 2006, 128, 7404. (411) Cheng, J. H.; Li, Z. W.; Haught, M.; Tang, Y. C. Chem. Commun. 2006, 4617. (412) Xu, Z.; Oxgaard, J.; Goddard, W. A., III Organometallics 2008, 27, 3770. (413) Vedernikov, A. N.; Fettinger, J. C.; Mohr, F. J. Am. Chem. Soc. 2004, 126, 11160. (414) Khaskin, E.; Zavalij, P. Y.; Vedernikov, A. N. J. Am. Chem. Soc. 2006, 128, 13054. (415) Muller, R. P.; Philipp, D. M.; Goddard, W. A., III Top. Catal. 2003, 23, 81. (416) Young, K. J. H.; Oxgaard, J.; Ess, D. H.; Meier, S. K.; Stewart, T.; Goddard, W. A., III; Periana, R. A. Chem. Commun. 2009, 3270.

Balcells et al. (417) Jones, C. J.; Taube, D.; Ziatdinov, V. R.; Periana, R. A.; Nielsen, R. J.; Oxgaard, J.; Goddard, W. A., III Angew. Chem., Int. Ed. 2004, 43, 4626. (418) Periana, R. A.; Mironov, O.; Taube, D.; Bhalla, G.; Jones, C. J. Science 2003, 301, 814. (419) Chempath, S.; Bell, A. T. J. Am. Chem. Soc. 2006, 128, 4650. (420) Lin, M.; Sen, A. Nature 1994, 368, 613. (421) Lin, M.; Hogan, T. E.; Sen, A. J. Am. Chem. Soc. 1996, 118, 4574. (422) Hristov, J. H.; Ziegler, T. Organometallics 2003, 22, 3513. (423) Strassner, T.; Houk, K. N. J. Am. Chem. Soc. 2000, 122, 7821. (424) Lam, W. W. Y.; Yiu, S.-M.; Lee, J. M. N.; Yau, S. K. Y.; Kwong, H.-K.; Lau, T.-C.; Liu, D.; Lin, Z. J. Am. Chem. Soc. 2006, 128, 2851. (425) Ensing, B.; Buda, F.; Gribnau, M. C. M.; Baerends, E. J. J. Am. Chem. Soc. 2004, 126, 4355. (426) Buda, F.; Ensing, B.; Gribnau, M. C. M.; Baerends, E. J. Chem.sEur. J. 2001, 7, 2775. (427) Ensing, B.; Buda, F.; Blochl, P. E.; Baerends, E. J. Phys. Chem. Chem. Phys. 2002, 4, 3619. (428) Louwerse, M. J.; Vassilev, P.; Baerends, E. J. J. Phys. Chem. A 2008, 112, 1000. (429) Bernasconi, L.; Louwerse, M. J.; Baerends, E. J. Eur. J. Inorg. Chem. 2007, 3023. (430) Bernasconi, L.; Baerends, E. J. Eur. J. Inorg. Chem. 2008, 1672. (431) Louwerse, M. J.; Baerends, E. J. Phys. Chem. Chem. Phys. 2007, 9, 156. (432) Michel, C.; Baerends, E. J. Inorg. Chem. 2009, 48, 3628. (433) Shiota, Y.; Kihara, N.; Kamachi, T.; Yoshizawa, K. J. Org. Chem. 2003, 68, 3958. (434) Costas, M.; Chen, K.; Que, L. Coord. Chem. ReV. 2000, 200, 517. (435) Bassan, A.; Blomberg, M. R. A.; Siegbahn, P. E. M.; Que, L. Chem.sEur. J. 2005, 11, 692. (436) Ma, Y.; Balbuena, P. B. J. Phys. Chem. B 2007, 111, 2711. (437) Kumar, D.; Hirao, H.; Que, L.; Shaik, S. J. Am. Chem. Soc. 2005, 127, 8026. (438) Hirao, H.; Kumar, D.; Que, L.; Shaik, S. J. Am. Chem. Soc. 2006, 128, 8590. (439) Schro¨der, D.; Shaik, S.; Schwarz, H. Acc. Chem. Res. 2000, 33, 139. (440) Shaik, S.; de Visser, S. P.; Ogliaro, F.; Schwarz, H.; Schro¨der, D. Curr. Opin. Chem. Biol. 2002, 6, 556. (441) Shaik, S.; Filatov, M.; Schro¨der, D.; Schwarz, H. Chem.sEur. J. 1998, 4, 193. (442) Sastri, C. V.; Lee, J.; Oh, K.; Lee, Y. J.; Lee, J.; Jackson, T. A.; Ray, K.; Hirao, H.; Shin, W.; Halfen, J. A.; Kim, J.; Que, L.; Shaik, S.; Nam, W. Proc. Natl. Acad. Sci. U.S.A. 2007, 104, 19181. (443) Hirao, H.; Que, L.; Nam, W.; Shaik, S. Chem.sEur. J. 2008, 14, 1740. (444) Dhuri, S. N.; Seo, M. S.; Lee, Y.-M.; Hirao, H.; Wang, Y.; Nam, W.; Shaik, S. Angew. Chem., Int. Ed. 2008, 47, 3356. (445) Klinker, E. J.; Shaik, S.; Hirao, H.; Que, L. Angew. Chem., Int. Ed. 2009, 48, 1291. (446) Hirao, H.; Chen, H.; Carvajal, M. A.; Wang, Y.; Shaik, S. J. Am. Chem. Soc. 2008, 130, 3319. (447) de Visser, S. P. J. Am. Chem. Soc. 2006, 128, 15809. (448) Decker, A.; Rohde, J.-U.; Klinker, E. J.; Wong, S. D.; Que, L.; Solomon, E. I. J. Am. Chem. Soc. 2007, 129, 15983. (449) Decker, A.; Solomon, E. I. Angew. Chem., Int. Ed. 2005, 44, 2252. (450) Borowski, T.; Bassan, A.; Siegbahn, P. E. M. Inorg. Chem. 2004, 43, 3277. (451) de Visser, S. P.; Oh, K.; Han, A.-R.; Nam, W. Inorg. Chem. 2007, 46, 4632. (452) Wang, Y.; Wang, Y.; Han, K. J. Biol. Inorg. Chem. 2009, 14, 533. (453) Klinker, E. J.; Brennessel, W. W.; Woodrum, N. L.; Cramer, C. J.; Que, L. Angew. Chem., Int. Ed. 2005, 44, 3690. (454) Noack, H.; Siegbahn, P. E. M. J. Biol. Inorg. Chem. 2007, 12, 1151. (455) Comba, P.; Maurer, M.; Vadivelu, P. J. Phys. Chem. A 2008, 112, 13028. (456) Johansson, A. J.; Blomberg, M. R. A.; Siegbahn, P. E. M. J. Phys. Chem. C 2007, 111, 12397. (457) Rhode, J. U.; Que, L. Angew. Chem., Int. Ed. 2005, 44, 2255. (458) de Visser, S. P. J. Am. Chem. Soc. 2006, 128, 9813. (459) de Visser, S. P. Angew. Chem., Int. Ed. 2006, 45, 1790. (460) Godfrey, E.; Porro, C. S.; de Visser, S. P. J. Phys. Chem. A 2008, 112, 2464. (461) Bassan, A.; Blomberg, M. R. A.; Siegbahn, P. E. M. Chem.sEur. J. 2003, 9, 4055. (462) Bassan, A.; Blomberg, M. R. A.; Siegbahn, P. E. M. Chem.sEur. J. 2003, 9, 106. (463) Shiota, Y.; Yoshizawa, K. J. Phys. Chem. B 2004, 108, 17226. (464) Borowski, T.; de Marothy, S.; Broclawik, E.; Schofield, C. J.; Siegbahn, P. E. M. Biochemistry 2007, 46, 3682. (465) Iwata-Reuyl, D.; Basak, A.; Townsend, C. J. Am. Chem. Soc. 1999, 121, 11356.

CsH Bond Activation in Transition Metal Species (466) Basak, A.; Salowe, S.; Townsend, C. J. Am. Chem. Soc. 1990, 112, 1654. (467) Wirstam, M.; Siegbahn, P. E. M. J. Am. Chem. Soc. 2000, 122, 8539. (468) Lundberg, M.; Siegbahn, P. E. M.; Morokuma, K. Biochemistry 2008, 47, 1031. (469) Liu, H.; Llano, J.; Gauld, J. W. J. Phys. Chem. B 2009, 113, 4887. (470) De Angelis, F.; Jin, N.; Car, R.; Groves, J. T. Inorg. Chem. 2006, 45, 4268. (471) Huber, S. M.; Ertem, M. Z.; Aquilante, F.; Gagliardi, L.; Tolman, W. B.; Cramer, C. J. Chem.sEur. J. 2009, 15, 4886. (472) Kamachi, T.; Kihara, N.; Shiota, Y.; Yoshizawa, K. Inorg. Chem. 2005, 44, 4226. (473) Yoshizawa, K.; Kihara, N.; Kamachi, T.; Shiota, Y. Inorg. Chem. 2006, 45, 3034. (474) Crespo, A.; Marti, M. A.; Roitberg, A. E.; Amzel, L. M.; Estrin, D. A. J. Am. Chem. Soc. 2006, 128, 12817. (475) Szaleniec, M.; Witko, M.; Heider, J. J. Mol. Catal. A: Chem. 2008, 286, 128. (476) Olsson, M. H. M.; Siegbahn, P. E. M.; Warshel, A. J. Am. Chem. Soc. 2004, 126, 2820. (477) Hatcher, E.; Soudackov, A. V.; Hammes-Schiffer, S. J. Am. Chem. Soc. 2004, 126, 5763. (478) Hatcher, E.; Soudackov, A. V.; Hammes-Schiffer, S. J. Am. Chem. Soc. 2007, 129, 187. (479) Karlsson, E. A.; Privalov, T. Chem.sEur. J. 2009, 15, 1862. (480) Owens, G. S.; Abu-Omar, M. M. Chem. Commun. 2000, 1165. (481) Bianchini, G.; Crucianelli, M.; De Angelis, F.; Neri, V.; Saladino, R. Tetrahedron Lett. 2005, 46, 2427. (482) Lehnert, N.; Neese, F.; Ho, R. Y. N.; Que, L.; Solomon, E. I. J. Am. Chem. Soc. 2002, 124, 10810. (483) Kumar, D.; Hirao, H.; Shaik, S.; Kozlowski, P. M. J. Am. Chem. Soc. 2006, 128, 16148. (484) Worth, L.; Frank, B. L.; Christner, D. F.; Absalon, M. J.; Stubbe, J.; Kozarich, J. W. Biochemistry 1993, 32, 2601. (485) Neese, F.; Zaleski, J. M.; Zaleski, K. L.; Solomon, E. I. J. Am. Chem. Soc. 2000, 122, 11703. (486) Decker, A.; Chow, M. S.; Kemsley, J. N.; Lehnert, N.; Solomon, E. I. J. Am. Chem. Soc. 2006, 128, 4719. (487) Chow, M. S.; Liu, L. V.; Solomon, E. I. Proc. Natl. Acad. Sci. U.S.A. 2008, 105, 13241. (488) Cramer, C. J.; Kinsinger, C. R.; Pak, Y. J. Mol. Struct. (Theochem) 2003, 632, 111. (489) Yoshizawa, K. Acc. Chem. Res. 2006, 39, 375. (490) Yoshizawa, K. Coord. Chem. ReV. 2002, 226, 251. (491) Yoshizawa, K. J. Biol. Inorg. Chem. 1998, 3, 318. (492) Torrent, M.; Musaev, D. G.; Basch, H.; Morokuma, K. J. Comput. Chem. 2002, 23, 59. (493) Siegbahn, P. E. M.; Blomberg, M. R. A. Chem. ReV. 2000, 100, 421. (494) Himo, F.; Siegbahn, P. E. M. Chem. ReV. 2003, 103, 2421. (495) Siegbahn, P. E. M.; Borowski, T. Acc. Chem. Res. 2006, 39, 729. (496) Siegbahn, P. E. M.; Crabtree, R. H. Struct. Bonding (Berlin) 2000, 97, 125. (497) Blomberg, M. R. A.; Siegbahn, P. E. M. J. Phys. Chem. B 2001, 105, 9375. (498) Friesner, R. A.; Dunietz, B. D. Acc. Chem. Res. 2001, 34, 351. (499) Guallar, V.; Gherman, B. F.; Lippard, S. J.; Friesner, R. A. Curr. Opin. Chem. Biol. 2002, 6, 236. (500) Friesner, R. A.; Baik, M. H.; Gherman, B. F.; Guallar, V.; Wirstam, M.; Murphy, R. B.; Lippard, S. J. Coord. Chem. ReV. 2003, 238, 267. (501) Baik, M. H.; Newcomb, M.; Friesner, R. A.; Lippard, S. J. Chem. ReV. 2003, 103, 2385. (502) Schro¨der, D.; Schwarz, H. Angew. Chem., Int. Ed. Engl. 1995, 34, 1973. (503) Schro¨der, D.; Schwarz, H.; Shaik, S. Struct. Bonding (Berlin) 2000, 97, 91. (504) Schro¨der, D.; Schwarz, H. Top Organomet. Chem. 2007, 22, 1. (505) Schro¨der, D.; Schwarz, H. Proc. Natl. Acad. Sci. U.S.A. 2008, 105, 18114. (506) Fiedler, A.; Schro¨der, D.; Shaik, S.; Schwarz, H. J. Am. Chem. Soc. 1994, 116, 10734. (507) Schro¨der, D.; Schwarz, H. Angew. Chem., Int. Ed. Engl. 1990, 29, 1433. (508) Shiota, Y.; Yoshizawa, K. J. Am. Chem. Soc. 2000, 122, 12317. (509) Yoshizawa, K.; Shiota, Y.; Yamabe, T. Chem.sEur. J. 1997, 3, 1160. (510) Yoshizawa, K.; Yamabe, T.; Hoffmann, R. New J. Chem. 1997, 21, 151. (511) Yoshizawa, K.; Shiota, Y.; Yamabe, T. J. Am. Chem. Soc. 1998, 120, 564. (512) Yoshizawa, K.; Shiota, Y.; Yamabe, T. Organometallics 1998, 17, 2825. (513) Yoshizawa, K.; Shiota, Y.; Kagawa, Y.; Yamabe, T. J. Phys. Chem. A 2000, 104, 2552. (514) Yoshizawa, K.; Shiota, Y.; Yamabe, T. J. Chem. Phys. 1999, 111, 538.

Chemical Reviews, 2010, Vol. 110, No. 2 823 (515) Yoshizawa, K.; Ohta, T.; Shiota, Y.; Yamabe, T. Chem. Lett. 1997, 1213. (516) Yoshizawa, K.; Ohta, T.; Yamabe, T.; Hoffmann, R. J. Am. Chem. Soc. 1997, 119, 12311. (517) Yoshizawa, K.; Ohta, T.; Yamabe, T. Bull. Chem. Soc. Jpn. 1998, 71, 1899. (518) Yoshizawa, K. J. Inorg. Biochem. 2000, 78, 23. (519) Yoshizawa, K.; Suzuki, A.; Shiota, Y.; Yamabe, T. Bull. Chem. Soc. Jpn. 2000, 73, 815. (520) Noodleman, L.; Lovell, T.; Han, W.-G.; Li, J.; Himo, F. Chem. ReV. 2004, 104, 459. (521) Yoshizawa, K.; Yumura, T. Chem.sEur. J. 2003, 92, 347. (522) Shu, L. J.; Nesheim, J. C.; Kauffmann, K.; Munck, E.; Lipscomb, J. D.; Que, L. Science 1997, 275, 515. (523) Yoshizawa, K.; Shiota, Y. J. Am. Chem. Soc. 2006, 128, 9873. (524) Shiota, Y.; Yoshizawa, K. Inorg. Chem. 2009, 48, 838. (525) Chen, P. P. Y.; Chan, S. I. J. Inorg. Biochem. 2006, 100, 801. (526) Chan, S. I.; Chen, K. H. C.; Yu, S. S. F.; Chen, C. L.; Kuo, S. S. J. Biochemistry 2004, 43, 4421. (527) Torrent, M.; Vreven, T.; Musaev, D. G.; Morokuma, K.; Farkas, O.; Schlegel, H. B. J. Am. Chem. Soc. 2002, 124, 192. (528) Basch, H.; Mogi, K.; Musaev, D. G.; Morokuma, K. J. Am. Chem. Soc. 1999, 121, 7249. (529) Basch, H.; Musaev, D. G.; Morokuma, K. J. Phys. Chem. B 2001, 105, 8452. (530) Basch, H.; Musaev, D. G.; Mogi, K.; Morokuma, K. J. Phys. Chem. A 2001, 105, 3615. (531) Musaev, D. G.; Basch, H.; Morokuma, K. J. Am. Chem. Soc. 2002, 124, 4135. (532) Huynh, M. H. V.; Meyer, T. J. Chem. ReV. 2007, 107, 5004. (533) DeWitt, J. G.; Bentsen, J. G.; Rosenzweig, A. C.; Hedman, B.; Green, J.; Pilkington, S.; Papaefthymiou, G. C.; Dalton, H.; Hodgson, K. O.; Lippard, S. J. J. Am. Chem. Soc. 1991, 113, 9219. (534) Fox, B. G.; Surerus, K. K.; Munck, E.; Lipscomb, J. D. J. Biol. Chem. 1988, 263, 10553. (535) Fox, B. G.; Hendrich, M. P.; Surerus, K. K.; Andersson, K. K.; Froland, W. A.; Lipscomb, J. D.; Munck, E. J. Am. Chem. Soc. 1993, 115, 3688. (536) Siegbahn, P. E. M.; Crabtree, R. H. J. Am. Chem. Soc. 1997, 119, 3103. (537) Siegbahn, P. E. M. Inorg. Chem. 1999, 38, 2880. (538) Siegbahn, P. E. M.; Crabtree, R. H.; Nordlund, P. J. Biol. Inorg. Chem. 1998, 3, 314. (539) Siegbahn, P. E. M. J. Biol. Inorg. Chem. 2001, 6, 27. (540) Dunietz, B. D.; Beachy, M. D.; Cao, Y. X.; Whittington, D. A.; Lippard, S. J.; Friesner, R. A. J. Am. Chem. Soc. 2000, 122, 2828. (541) Lovell, T.; Li, J.; Noodleman, L. Inorg. Chem. 2001, 40, 5251. (542) Lovell, T.; Li, J.; Noodleman, L. Inorg. Chem. 2001, 40, 5267. (543) Lovell, T.; Li, J.; Noodleman, L. J. Biol. Inorg. Chem. 2002, 7, 799. (544) Rosenzweig, A. C.; Nordlund, P.; Takahara, P. M.; Frederick, C. A.; Lippard, S. J. Chem. Biol. 1995, 2, 409. (545) Gherman, B. F.; Dunietz, B. D.; Whittington, D. A.; Lippard, S. J.; Friesner, R. A. J. Am. Chem. Soc. 2001, 123, 3836. (546) Valentine, A. M.; Wilkinson, B.; Liu, K. E.; Komar-Panicucci, S.; Priestley, N. D.; Williams, P. G.; Morimoto, H.; Floss, H. G.; Lippard, S. J. J. Am. Chem. Soc. 1997, 119, 1818. (547) Priestley, N. D.; Floss, H. G.; Froland, W. A.; Lipscomb, J. D.; Williams, P. G.; Morimoto, H. J. Am. Chem. Soc. 1992, 114, 7561. (548) Guallar, V.; Gherman, B. F.; Miller, W. H.; Lippard, S. J.; Friesner, R. A. J. Am. Chem. Soc. 2002, 124, 3377. (549) Baik, M. H.; Gherman, B. F.; Friesner, R. A.; Lippard, S. J. J. Am. Chem. Soc. 2002, 124, 14608. (550) Gherman, B. F.; Lippard, S. J.; Friesner, R. A. J. Am. Chem. Soc. 2005, 127, 1025. (551) Ambundo, E. A.; Friesner, R. A.; Lippard, S. J. J. Am. Chem. Soc. 2002, 124, 8770. (552) Ito, N.; Phillips, S. E. V.; Stevens, C.; Ogel, Z. B.; McPherson, M. J.; Keen, J. N.; Yadav, K. D. S.; Knowles, P. F. Nature 1991, 350, 87. (553) Branchaud, B. P.; Montague-Smith, M. P.; Kosman, D. J.; McLaren, F. R. J. Am. Chem. Soc. 1993, 115, 798. (554) Rothlisberger, U.; Carloni, P. Int. J. Quantum Chem. 1999, 73, 209. (555) Rothlisberger, U.; Carloni, P.; Doclo, K.; Parrinello, M. J. Biol. Inorg. Chem. 2000, 5, 236. (556) Guidoni, L.; Spiegel, K.; Zumstein, M.; Rothlisberger, U. Angew. Chem., Int. Ed. 2004, 43, 3286. (557) Zueva, E.; Walton, P. H.; McGrady, J. E. Dalton Trans. 2006, 159. (558) Bachler, V.; Chaudhuri, P.; Wieghardt, K. Chem.sEur. J. 2001, 7, 404. (559) Himo, F.; Eriksson, L. A.; Maseras, F.; Siegbahn, P. E. M. J. Am. Chem. Soc. 2000, 122, 8031. (560) Cheng, L.; Wang, J.; Wang, M.; Wu, Z. Dalton Trans. 2009, 3286.

CR900315K